mirror of
https://github.com/garrytan/gstack.git
synced 2026-11-06 17:04:27 +01:00
fix: recognize native review narration and controls
This commit is contained in:
Garry Tan committed
1 parent
d4f42b69a1
commit
7fac51e9bd
17 files changed
+1308
-59
No files matched your search
@@ -0,0 +1,134 @@
|
||||
import {expect,test} from 'bun:test';
|
||||
import fs from 'node:fs';
|
||||
import os from 'node:os';
|
||||
import path from 'node:path';
|
||||
import {readPlanCountTranscript,type NativePublicToolEvent} from './helpers/plan-count-transcript';
|
||||
import {autoplanPhaseCompletions} from './helpers/autoplan-phase-observer';
|
||||
import fixture from './fixtures/autoplan-public-narration-ad.json';
|
||||
import {E2E_TOUCHFILES,selectTests} from './helpers/touchfiles';
|
||||
|
||||
const at=Date.parse(fixture.provenance.timestamp);
|
||||
function read(blocks: unknown[]= [fixture.block],delta: any={},complete=true) {
|
||||
const dir=fs.mkdtempSync(path.join(os.tmpdir(),'public-narration-')),cwd=path.join(dir,'repo');
|
||||
const project=path.join(dir,'projects','owned');fs.mkdirSync(project,{recursive:true});
|
||||
const record={cwd,isSidechain:false,sessionId:fixture.provenance.sessionId,
|
||||
timestamp:fixture.provenance.timestamp,uuid:fixture.provenance.uuid,
|
||||
message:{role:'assistant',content:blocks},...delta};
|
||||
const file=path.join(project,fixture.provenance.sessionId+'.jsonl');
|
||||
fs.writeFileSync(file,JSON.stringify(record)+(complete?'\n':''));
|
||||
const tools:NativePublicToolEvent[]=[];
|
||||
try{return {transcript:readPlanCountTranscript(dir,cwd,event=>tools.push(event)),tools};}
|
||||
finally{fs.rmSync(dir,{recursive:true,force:true});}
|
||||
}
|
||||
|
||||
test('actual public server narration is projected at its original session and timestamp',()=>{
|
||||
const {transcript,tools}=read();
|
||||
expect(transcript.assistantMessages).toEqual([{sessionId:fixture.provenance.sessionId,
|
||||
timestamp:fixture.provenance.timestamp,text:fixture.block.thinking}]);
|
||||
expect(transcript.calls).toEqual([]);expect(tools).toEqual([]);
|
||||
expect(autoplanPhaseCompletions(transcript,at-1)).toEqual([{phase:1,ts:at}]);
|
||||
});
|
||||
test('actual Phase 1 is done declaration independently matches the phase marker',()=>{
|
||||
expect(autoplanPhaseCompletions({status:'ready',calls:[],assistantMessages:[{
|
||||
sessionId:fixture.provenance.sessionId,timestamp:fixture.provenance.timestamp,
|
||||
text:fixture.block.thinking}]},at-1)).toEqual([{phase:1,ts:at}]);
|
||||
});
|
||||
|
||||
// Minimal synthetic protobuf envelopes exercise public-tag classification only.
|
||||
// No opaque native signature or untagged model text is stored in this fixture.
|
||||
const vi=(n:number):number[]=>{const bytes:number[]=[];do{const b=n%128;n=Math.floor(n/128);bytes.push(b+(n?128:0));}while(n);return bytes;};
|
||||
const field=(n:number,body:Uint8Array)=>Buffer.from([...vi(n*8+2),...vi(body.length),...body]);
|
||||
const tagged=(kind='narration')=>field(2,field(1,field(8,Buffer.from(kind))));
|
||||
const summary=(signature=fixture.block.signature,thinking=fixture.block.thinking)=>({type:'thinking',thinking,signature});
|
||||
const encode=(bytes:Uint8Array)=>Buffer.from(bytes).toString('base64');
|
||||
|
||||
test('unknown or malformed signature envelopes never become public narration',()=>{
|
||||
const invalid:unknown[]=[undefined,null,'',4,'narration','%%%',' '+fixture.block.signature,
|
||||
fixture.block.signature+'=',encode(tagged('reasoning')),encode(tagged('Narration')),
|
||||
encode(field(1,field(1,field(8,Buffer.from('narration'))))),
|
||||
encode(field(2,field(2,field(8,Buffer.from('narration'))))),
|
||||
encode(field(2,field(1,field(7,Buffer.from('narration'))))),
|
||||
encode(tagged().subarray(0,-1)),encode(Buffer.from([0x12,0x80])),
|
||||
encode(Buffer.from([0x12,0xff,0xff,0xff,0xff,0xff,0xff,0xff,0xff,0x7f])),
|
||||
encode(Buffer.concat([tagged(),Buffer.from([0])])),
|
||||
encode(Buffer.concat([tagged(),Buffer.from([0x0f])])),
|
||||
encode(Buffer.concat([tagged(),tagged('reasoning')])),
|
||||
encode(field(2,Buffer.concat([field(1,field(8,Buffer.from('narration'))),field(1,field(8,Buffer.from('reasoning')))]))),
|
||||
encode(field(2,field(1,Buffer.concat([field(8,Buffer.from('narration')),field(8,Buffer.from('reasoning'))])))),
|
||||
encode(field(2,field(1,Buffer.concat([field(8,Buffer.from('narration')),field(8,Buffer.from('narration'))])))),
|
||||
encode(Buffer.concat([tagged(),field(3,Buffer.alloc(64*1024))])),
|
||||
];
|
||||
for(const signature of invalid){
|
||||
const {transcript,tools}=read([{...summary(),signature}]);
|
||||
expect(transcript.assistantMessages,JSON.stringify(signature)?.slice(0,80)).toEqual([]);
|
||||
expect(transcript.calls).toEqual([]);expect(tools).toEqual([]);
|
||||
}
|
||||
});
|
||||
test('supported unrelated envelope fields preserve an exact public tag',()=>{
|
||||
const bytes=Buffer.concat([Buffer.from([0x08,0x01]),tagged(),field(3,Buffer.from('opaque'))]);
|
||||
expect(read([summary(encode(bytes))]).transcript.assistantMessages[0]?.text).toBe(fixture.block.thinking);
|
||||
});
|
||||
test('private, empty and wrong-kind blocks remain outside the public projection',()=>{
|
||||
for(const block of [{type:'thinking',thinking:'SYNTHETIC_PRIVATE_TEXT'},
|
||||
{...summary(),signature:encode(tagged('reasoning')),thinking:'SYNTHETIC_PRIVATE_TEXT'},
|
||||
{...summary(),thinking:''},{...summary(),thinking:' '},{...summary(),thinking:7},
|
||||
{...summary(),type:'redacted_thinking'},{...summary(),type:'tool_result'},
|
||||
{type:'thinking',thinking:'SYNTHETIC_PRIVATE_TEXT',block_kind:'narration'}]){
|
||||
expect(read([block]).transcript.assistantMessages).toEqual([]);
|
||||
}
|
||||
});
|
||||
test('parent role, cwd, native filename and complete valid timestamp remain required',()=>{
|
||||
for(const delta of [{cwd:'/foreign'},{sessionId:'foreign'},
|
||||
{isSidechain:true},{isSidechain:undefined},{timestamp:'invalid'},
|
||||
{timestamp:null},{message:{role:'user',content:[fixture.block]}},
|
||||
{message:{role:'system',content:[fixture.block]}}]){
|
||||
expect(read([fixture.block],delta).transcript.assistantMessages).toEqual([]);
|
||||
}
|
||||
expect(read([fixture.block],{},false).transcript.assistantMessages).toEqual([]);
|
||||
});
|
||||
test('public narration does not manufacture questions, replies or plan approval',()=>{
|
||||
const {transcript,tools}=read([summary(),{type:'text',text:'Ordinary assistant prose.'},
|
||||
{type:'thinking',thinking:'SYNTHETIC_PRIVATE_TEXT'},
|
||||
{type:'tool_use',id:'actual-read',name:'Read',input:{file_path:'/owned/PLAN.md'}}]);
|
||||
expect(transcript.assistantMessages.map(m=>m.text)).toEqual([fixture.block.thinking,'Ordinary assistant prose.']);
|
||||
expect(transcript.calls).toEqual([]);expect(transcript.planReadyRequests).toBeUndefined();
|
||||
expect(tools.map(e=>[e.kind,e.toolUseId,e.name])).toEqual([['use','actual-read','Read']]);
|
||||
});
|
||||
function hits(text:string,start=at-1,timestamp=fixture.provenance.timestamp){
|
||||
return autoplanPhaseCompletions({status:'ready',calls:[],assistantMessages:[{
|
||||
sessionId:fixture.provenance.sessionId,timestamp,text}]},start);
|
||||
}
|
||||
test('new done declarations retain exact phase numbers, punctuation and timestamps',()=>{
|
||||
for(const phase of [1,2,2.5,3])for(const tail of ['', '.', ': Work retained.', '. Work retained.']){
|
||||
expect(hits(`Phase ${phase} is done${tail}`)).toEqual([{phase,ts:at}]);
|
||||
}
|
||||
expect(hits('**Phase 1 is done.**')).toEqual([{phase:1,ts:at}]);
|
||||
expect(hits(fixture.block.thinking,at+1)).toEqual([]);
|
||||
expect(hits(fixture.block.thinking,at-1,'invalid')).toEqual([]);
|
||||
expect(hits('Phase 1 complete. Work retained.')).toEqual([{phase:1,ts:at}]);
|
||||
});
|
||||
test('source examples, questions, promises and quoted done markers are not phase completion',()=>{
|
||||
for(const text of ['Phase 1 is done?', 'Phase 1 is done eventually', 'Phase 1 is not done.',
|
||||
'Once Phase 1 is done, continue.', 'Phase 1 will be done.', 'Phase 4 is done.',
|
||||
'Phase 2.1 is done.', '# Phase 1 is done.', '> Phase 1 is done.',
|
||||
'- Phase 1 is done.', '| Phase 1 is done. |', ' Phase 1 is done.',
|
||||
'```text\nPhase 1 is done.\n```', '~~~text\nPhase 1 is done.\n~~~',
|
||||
'Example:\nPhase 1 is done.\nPhase 2 is done.',
|
||||
'Emit phase-transition summary: Phase 1 is done.',
|
||||
'**Phase 1 is done** if the tests pass.']) expect(hits(text),text).toEqual([]);
|
||||
});
|
||||
test('phase ordering and duplicate collapse use native time rather than polling order',()=>{
|
||||
const t={status:'ready' as const,calls:[],assistantMessages:[
|
||||
{sessionId:'parent',timestamp:new Date(at+20).toISOString(),text:'Phase 2 is done.'},
|
||||
{sessionId:'parent',timestamp:new Date(at+10).toISOString(),text:fixture.block.thinking},
|
||||
{sessionId:'parent',timestamp:new Date(at+30).toISOString(),text:'Phase 1 is done.'}]};
|
||||
expect(autoplanPhaseCompletions(t,at)).toEqual([{phase:1,ts:at+10},{phase:2,ts:at+20}]);
|
||||
});
|
||||
|
||||
test('public narration changes select every existing shared native-reader consumer',()=>{
|
||||
const reader=selectTests(['test/helpers/plan-count-transcript.ts'],E2E_TOUCHFILES).selected.sort();
|
||||
expect(reader).toHaveLength(8);expect(reader).toContain('autoplan-chain-pty');
|
||||
expect(reader).toContain('plan-ceo-mode-routing');
|
||||
for(const file of ['test/autoplan-public-narration.test.ts','test/fixtures/autoplan-public-narration-ad.json'])
|
||||
expect(selectTests([file],E2E_TOUCHFILES).selected.sort()).toEqual(reader);
|
||||
});
|
||||
@@ -0,0 +1,119 @@
|
||||
import {describe,expect,test} from 'bun:test';
|
||||
import fs from 'node:fs';import os from 'node:os';import path from 'node:path';
|
||||
import {hasNativePostAnswerCeoPosture,nextCeoModeNavigation} from './helpers/ceo-mode-option';
|
||||
import {capturePlanCountQuestion,nativePlanCallFingerprint,planCountPrerequisitePick,planCountQuestionInput} from './helpers/claude-pty-runner';
|
||||
import {readPlanCountTranscript,type NativePublicToolEvent,type NativePlanQuestionCall} from './helpers/plan-count-transcript';
|
||||
import captured from './fixtures/ceo-mode-full-ad.json';
|
||||
import {E2E_TOUCHFILES,selectTests} from './helpers/touchfiles';
|
||||
const pattern=/\b(expansion|10x|delight|dream|cathedral|opt[\s-]?in)\b/i;
|
||||
function replay(i:number){
|
||||
const item=captured.cases[i]!,root=fs.mkdtempSync(path.join(os.tmpdir(),'ceo-full-ad-'));
|
||||
fs.mkdirSync(path.join(root,'projects','owned'),{recursive:true});
|
||||
fs.writeFileSync(path.join(root,'projects','owned',item.process.sessionId+'.jsonl'),item.records.map(r=>JSON.stringify(r)).join('\n')+'\n');
|
||||
const events:NativePublicToolEvent[]=[];
|
||||
try{return {item,transcript:readPlanCountTranscript(root,item.process.cwd,e=>events.push(e)),events};}
|
||||
finally{fs.rmSync(root,{recursive:true,force:true});}
|
||||
}
|
||||
function pending(){const c=structuredClone(replay(0).transcript.calls[0]!);c.answered=false;delete c.answers;delete c.answeredAt;delete c.unansweredQuestionIndices;return c;}
|
||||
// Full panes projected from exact native questions, not retained historical viewports.
|
||||
function pane(call:NativePlanQuestionCall,index:number){const q=call.questions[index]!;return [
|
||||
call.questions.length>1?'← '+call.questions.map((v,i)=>`${i<index?'☒':'☐'} ${v.header}`).join(' ')+' ✔ Submit →':'☐ '+q.header,
|
||||
q.question,...q.options.map((v,i)=>`${i?' ':'❯'} ${i+1}. ${v.label}`),
|
||||
`Enter to select · ${call.questions.length>1?'Tab/Arrow keys':'↑/↓'} to navigate · Esc to cancel`].join('\n');}
|
||||
function frame(c:NativePlanQuestionCall,index:number){const visible=pane(c,index);return {visible,active:capturePlanCountQuestion(visible,new Set(),0,true,c)!,routing:nativePlanCallFingerprint(c,0,true)};}
|
||||
function match(e= replay(1)){return hasNativePostAnswerCeoPosture(e.transcript,'SCOPE EXPANSION',pattern,e.item.selectedAt!,e.events);}
|
||||
function rebind(e:ReturnType<typeof replay>){const d=e.transcript.calls[1]!,q=d.questions[0]!;e.events[2]!.input={questions:d.questions};d.answers={[q.question]:q.options[0]!.label};}
|
||||
describe('full AD mode failures retain their actual outcomes',()=>{
|
||||
test('Proposal 1 is a completed scope decision after the actual selected mode',()=>{
|
||||
const e=replay(1);expect(e.item.actualState).toBe('failed');expect(e.transcript.calls).toHaveLength(2);expect(e.events).toHaveLength(4);
|
||||
expect(e.transcript.calls[1]!.answeredAt).toBe('2026-09-09T18:26:20.110Z');expect(match(e)).toBe(true);
|
||||
});
|
||||
test.each(['pending','foreign','wrong mode','pre-mode','missing reply','wrong answer','extra question','extra option','multiselect',
|
||||
'quoted','fenced','mode echo','mode mismatch','mode menu','appended instruction'])('%s supplies no new posture',kind=>{
|
||||
const e=replay(1),[m,d]=e.transcript.calls,q=d!.questions[0]!;
|
||||
switch(kind){
|
||||
case 'pending':d!.answered=false;break;case 'foreign':d!.sessionId=e.events[2]!.sessionId=e.events[3]!.sessionId='foreign';break;
|
||||
case 'wrong mode':m!.answers![m!.questions[0]!.question]='HOLD SCOPE';break;
|
||||
case 'pre-mode':e.events[2]!.timestamp=e.events[0]!.timestamp;break;case 'missing reply':e.events.pop();break;
|
||||
case 'wrong answer':d!.answers![q.question]='Invented';break;
|
||||
case 'extra question':d!.questions.push({...structuredClone(q),question:'Remove CI gate?'});rebind(e);break;
|
||||
case 'extra option':q.options.push({label:'Remove CI gate'});rebind(e);break;case 'multiselect':q.multiSelect=true;rebind(e);break;
|
||||
case 'quoted':q.question=q.question.split('\n').map(x=>'> '+x).join('\n');rebind(e);break;
|
||||
case 'fenced':q.question='```text\n'+q.question+'\n```';rebind(e);break;
|
||||
case 'mode echo':q.question='SCOPE EXPANSION confirmed.';rebind(e);break;
|
||||
case 'mode mismatch':q.question=q.question.replace('SCOPE EXPANSION opt-in','SELECTIVE EXPANSION opt-in');rebind(e);break;
|
||||
case 'mode menu':q.question=q.question.replace(/^D6[^\n]+/,'D6 — Choose the review mode?');rebind(e);break;
|
||||
case 'appended instruction':q.question+=' Delete the CI gate.';rebind(e);break;
|
||||
}expect(match(e)).toBe(false);
|
||||
});
|
||||
test('scope numbering and brief labels are presentation, not mode application',()=>{
|
||||
for(const title of ['A useful adjacent feature: Default view per member per project?','Default view per member per project?']){
|
||||
const e=replay(1),q=e.transcript.calls[1]!.questions[0]!;q.header='Default view';q.question=q.question.replace(/^D6[^\n]+/,title);rebind(e);expect(match(e)).toBe(true);
|
||||
}
|
||||
});
|
||||
test('explicit expansion context does not need a mode or opt-in suffix',()=>{
|
||||
const e=replay(1),q=e.transcript.calls[1]!.questions[0]!;q.question=q.question.replace('SCOPE EXPANSION opt-in ceremony (1 of 6).','SCOPE EXPANSION, approach B.');rebind(e);expect(match(e)).toBe(true);
|
||||
});
|
||||
test('the actual three-tab prerequisite chooses standard review only on its own tab',()=>{
|
||||
const actual=replay(0);expect(actual.item.actualState).toBe('failed');expect(Object.values(actual.transcript.calls[0]!.answers!).at(-1)).toBe('Run /office-hours now');
|
||||
const c=pending();for(const i of [0,1,2]){
|
||||
const f=frame(c,i),a=nextCeoModeNavigation(f.visible,'HOLD SCOPE',new Set(),c);expect(a.kind).toBe('question');
|
||||
if(a.kind==='question'){expect(a.question.nativeQuestionIndex).toBe(i);expect(planCountQuestionInput(f.visible,a.question,a.index)).toBe(i===2?'2':'1');}
|
||||
expect(planCountPrerequisitePick(f.routing,f.active)).toBe(i===2?2:null);
|
||||
}
|
||||
});
|
||||
test('single and reordered native prerequisite tabs preserve the meaning of the skip',()=>{
|
||||
const c=pending();c.questions=[c.questions[2]!];let f=frame(c,0);expect(planCountPrerequisitePick(f.routing,f.active)).toBe(2);
|
||||
c.questions[0]!.options.reverse();f=frame(c,0);expect(planCountPrerequisitePick(f.routing,f.active)).toBe(1);
|
||||
});
|
||||
test.each(['wrong tab','wrong signature','wrong body','wrong order','no metadata','completed','failed','extra action','multiselect','conditional','extra remedy','no description'])('a %s cannot borrow the prerequisite action',kind=>{
|
||||
const c=pending();if(kind==='completed')c.answered=true;if(kind==='failed')c.failed=true;
|
||||
if(kind==='extra action')c.questions[2]!.options.push({label:'Accept risk'});
|
||||
if(kind==='multiselect')c.questions[2]!.multiSelect=true;
|
||||
if(kind==='conditional')c.questions[2]!.options[1]!.description+=' if all tests pass.';
|
||||
if(kind==='extra remedy')c.questions[2]!.options[1]!.description+=' Remove the CI gate.';
|
||||
if(kind==='no description')c.questions[2]!.options[1]!.description='';
|
||||
const f=frame(c,2);let a=f.active;
|
||||
if(kind==='wrong tab')a={...a,nativeQuestionIndex:0};if(kind==='wrong signature')a={...a,signature:'foreign:tool:question:2'};
|
||||
if(kind==='wrong body')a={...a,promptSnippet:'Choose a product direction.'};if(kind==='wrong order')a={...a,options:[...a.options].reverse()};
|
||||
if(kind==='no metadata')a={...a,nativeCall:undefined};
|
||||
expect(planCountPrerequisitePick(f.routing,a)).toBeNull();
|
||||
});
|
||||
});
|
||||
|
||||
describe('full AD HOLD retry completed sequencing rationale',()=>{
|
||||
function hold(){const e=replay(2);return {e,decision:e.transcript.calls[2]!,q:e.transcript.calls[2]!.questions[0]!};}
|
||||
function matches(e:ReturnType<typeof replay>){return hasNativePostAnswerCeoPosture(e.transcript,'HOLD SCOPE',/\b(rigor|bulletproof|hold\s*scope|maximum\s+rigor)\b/i,e.item.selectedAt!,e.events);}
|
||||
function bind(e:ReturnType<typeof replay>){const d=e.transcript.calls[2]!,q=d.questions[0]!;e.events[4]!.input={questions:d.questions};d.answers={[q.question]:q.options[0]!.label};}
|
||||
test('the actual completed rationale applies HOLD to work in the previously approved approach',()=>{
|
||||
const {e,decision,q}=hold();expect(e.item.actualState).toBe('failed');expect(e.transcript.calls).toHaveLength(3);
|
||||
const approach=e.transcript.calls[0]!;expect(Object.values(approach.answers!)).toEqual(['B: ViewState schema (recommended)']);
|
||||
expect(approach.questions[0]!.options[0]!.description).toContain('URL params');
|
||||
expect(decision.answeredAt).toBe('2026-09-09T18:35:05.273Z');expect(q.question).toContain('not new scope either way');expect(matches(e)).toBe(true);
|
||||
});
|
||||
test('three and four alternatives still express one completed review decision',()=>{
|
||||
for(const count of [3,4]){const {e,q}=hold();q.options.push({label:'Gate URL sync for the pilot'});if(count===4)q.options.push({label:'Run a limited URL sync pilot'});bind(e);expect(matches(e)).toBe(true);}
|
||||
});
|
||||
test.each(['pending','foreign','before mode','missing reply','failed reply','wrong answer','metadata only','bare echo','other mode',
|
||||
'quoted rationale','fenced rationale','duplicate options','extra question','extra instruction','multiselect'])('%s is not completed HOLD rationale',kind=>{
|
||||
const {e,decision,q}=hold();
|
||||
switch(kind){
|
||||
case 'pending':decision.answered=false;break;case 'foreign':decision.sessionId=e.events[4]!.sessionId=e.events[5]!.sessionId='foreign';break;
|
||||
case 'before mode':e.events[4]!.timestamp=e.events[0]!.timestamp;break;case 'missing reply':e.events.pop();break;case 'failed reply':e.events[5]!.isError=true;break;
|
||||
case 'wrong answer':decision.answers![q.question]='Invented';break;
|
||||
case 'metadata only':q.question=q.question.replace(/ELI10:[\s\S]*?\nStakes/,'ELI10: We will implement the URL codec.\nStakes');bind(e);break;
|
||||
case 'bare echo':q.question=q.question.replace(/ELI10:[\s\S]*?\nStakes/,'ELI10: HOLD SCOPE confirmed.\nStakes');bind(e);break;
|
||||
case 'other mode':q.question=q.question.replace(/HOLD SCOPE/g,'SCOPE EXPANSION');bind(e);break;
|
||||
case 'quoted rationale':q.question=q.question.replace('ELI10: Approach','ELI10:\n> Approach');bind(e);break;
|
||||
case 'fenced rationale':q.question=q.question.replace('ELI10: Approach','ELI10: ```Approach');bind(e);break;
|
||||
case 'duplicate options':q.options[1]!.label=q.options[0]!.label;bind(e);break;
|
||||
case 'extra question':decision.questions.push({...structuredClone(q),question:'Remove CI?'});bind(e);break;
|
||||
case 'extra instruction':q.question+=' Disable authentication.';bind(e);break;
|
||||
case 'multiselect':q.multiSelect=true;bind(e);break;
|
||||
}expect(matches(e)).toBe(false);
|
||||
});
|
||||
});
|
||||
|
||||
test('the exact full AD regressions select their periodic caller',()=>{
|
||||
for(const file of ['test/ceo-mode-full-ad.test.ts','test/fixtures/ceo-mode-full-ad.json']) expect(selectTests([file],E2E_TOUCHFILES,[]).selected).toEqual(['plan-ceo-mode-routing']);
|
||||
});
|
||||
@@ -102,8 +102,9 @@ test('HOLD requires the explicit out-of-scope deferral and its selected defer an
|
||||
e.transcript.calls[1]!.answers={[q.question]:q.options[1]!.label};expect(matches(e)).toBe(false);
|
||||
});
|
||||
|
||||
test('E-number shorthand requires matching header, add decision and explicit expansion context',()=>{
|
||||
for(const change of [(q:any)=>{q.header='E2 shared';},(q:any)=>{q.question=q.question.replace('E1: Add','E1: Mention');},
|
||||
test('completed expansion decisions require application of the selected mode',()=>{
|
||||
for(const change of [(q:any)=>{q.question='D6 — Continue the review?\nSelected SCOPE EXPANSION.';},
|
||||
(q:any)=>{q.question=q.question.replace('SCOPE EXPANSION mode','SELECTIVE EXPANSION mode');},
|
||||
(q:any)=>{q.question=q.question.replace('SCOPE EXPANSION mode','HOLD SCOPE mode');},
|
||||
(q:any)=>{q.options[1].label='Enable telemetry';}]){
|
||||
const e=replay(1);change(e.transcript.calls[1]!.questions[0]);rebind(e);expect(matches(e)).toBe(false);
|
||||
|
||||
@@ -10,7 +10,9 @@ import type { NativePlanQuestionCall } from './helpers/plan-count-transcript';
|
||||
import { E2E_TOUCHFILES, selectTests } from './helpers/touchfiles';
|
||||
import priorCalls from './fixtures/ceo-mode-prerequisite-o-calls.json';
|
||||
import directProceedCall from './fixtures/ceo-mode-prerequisite-q-call.json';
|
||||
const calls = [...priorCalls, directProceedCall];
|
||||
import fullAd from './fixtures/ceo-mode-full-ad.json';
|
||||
const fullAdQuestions=fullAd.cases[0]!.records.find(row=>row.message.role==='assistant')!.message.content[0]!.input.questions;
|
||||
const calls = [...priorCalls, directProceedCall, {call:{sessionId:'full-ad-projected',toolUseId:'full-ad-prerequisite',questions:fullAdQuestions,answered:false,failed:false}}];
|
||||
function ownedIdentity(pid: number): string | null {
|
||||
try { return execFileSync('ps', ['-p', String(pid), '-o', 'lstart=', '-o', 'command='], { encoding: 'utf8', timeout: 5000 }).trim(); }
|
||||
catch { return null; }
|
||||
@@ -90,7 +92,7 @@ describe('CEO mode prerequisite navigation', () => {
|
||||
for(const file of ['test/ceo-mode-prerequisite.test.ts','test/fixtures/ceo-mode-prerequisite-o-calls.json','test/fixtures/ceo-mode-prerequisite-q-call.json'])expect(selectTests([file],E2E_TOUCHFILES).selected).toEqual(['plan-ceo-mode-routing']);
|
||||
});
|
||||
});
|
||||
for(const fixtureIndex of [1,2,3])test.skipIf(process.platform==='win32')(`fake native PTY skips prerequisite ${fixtureIndex} and confirms target posture`,async()=>{
|
||||
for(const fixtureIndex of [1,2,3,4])test.skipIf(process.platform==='win32')(`fake native PTY skips prerequisite ${fixtureIndex} and confirms target posture`,async()=>{
|
||||
const dir=fs.mkdtempSync(path.join(os.tmpdir(),'ceo-mode-prerequisite-')),fake=path.join(dir,'fake-claude'),events=path.join(dir,'events.jsonl'),worker=path.join(dir,'worker.ts');
|
||||
const root=path.resolve(import.meta.dir,'..'),call=pending(fixtureIndex),screens=call.questions.map((_,i)=>pane(call,i)),expected=call.questions.map(q=>q.header==='Design doc'?'2':'1');
|
||||
fs.writeFileSync(fake,`#!${process.execPath}\n`+`
|
||||
|
||||
+18
@@ -0,0 +1,18 @@
|
||||
{
|
||||
"provenance": {
|
||||
"kind": "Retained public narration text; synthetic schema-only signature envelope",
|
||||
"label": "ship-source-ad-full-paid-20260909-v1",
|
||||
"sessionId": "8f4d4357-07f8-45bb-98b4-0c3d94e2e2a0",
|
||||
"uuid": "15ec6b49-9b23-4f07-9d4c-4e4265255ed5",
|
||||
"timestamp": "2026-09-09T18:31:15.976Z",
|
||||
"textSha256": "65a0f544c2031cbd86ee5afe366aadcb86887448cbfb35d8cef604c1b14e96ab",
|
||||
"attributionReceiptSha256": "42b71fb70382755268d811ebd7293ba36fa02a997668c2897a1761b9f0668a14",
|
||||
"frozenCliSha256": "26d020351e8112f4006790f3cfce43b4c9df0c1bb1d0e542364d64151b81d5ba",
|
||||
"signature": "Synthetic protobuf tag path 2\u21921\u21928=narration. Original opaque signature and all non-narration thinking text are excluded. This tests classification, not cryptographic verification."
|
||||
},
|
||||
"block": {
|
||||
"type": "thinking",
|
||||
"thinking": "Phase 1 is done: the Claude subagent found 14 issues (3 critical, 5 high, 6 medium), with 6 taste disagreements and 1 premise flagged for the gate. I've updated the implementation plan with 10 tasks and saved the CEO plan, and I'm now moving into Phase 2 (Design, since UI scope was detected).",
|
||||
"signature": "Eg0KC0IJbmFycmF0aW9u"
|
||||
}
|
||||
}
|
||||
Vendored
+665
@@ -0,0 +1,665 @@
|
||||
{
|
||||
"sourceHead": "d4f42b69a1dd0eb48d89500e03fb1c4407229284",
|
||||
"runId": "ship-source-ad-full-paid-20260909-v1",
|
||||
"capturedAt": "2026-09-09T18:32:49.223242+00:00",
|
||||
"provenance": "Exact retained public native AUQ request/reply rows only. First attempts naturally failed; new retries do not amend these outcomes. No private reasoning included. Pending navigation panes in tests are projected from these exact request questions; no claim that an intermediate full pane was retained.",
|
||||
"cases": [
|
||||
{
|
||||
"targetMode": "HOLD SCOPE",
|
||||
"attempt": "plan-ceo-review-mode-1788978153186-tPgIwZ",
|
||||
"actualState": "failed",
|
||||
"selectedAt": null,
|
||||
"commandStartedAt": 1788978151180,
|
||||
"pendingQuestionRecorder": {
|
||||
"status": "pending"
|
||||
},
|
||||
"process": {
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-MPwj22",
|
||||
"sessionId": "0956daf4-4c61-4aa4-b163-2c429430386b"
|
||||
},
|
||||
"observation": {
|
||||
"path": ".context/ship-source-ad-full-paid-20260909-v1/evals/job-7/shards/skill-e2e-plan-ceo-mode-routing/pty-count/ship-source-ad-full-paid-20260909-v1-7/plan-ceo-review-mode-1788978153186-tPgIwZ/observation.json",
|
||||
"sha256": "6b98133508077d50c292eebb0281df1a761569e73342d0a78dbf6bb6883f8a2a"
|
||||
},
|
||||
"native": {
|
||||
"captureKey": "native:1349450-870044:0956daf4-4c61-4aa4-b163-2c429430386b:119454070",
|
||||
"path": "/home/vercel-sandbox/gstack/.context/ship-source-ad-full-paid-20260909-v1/full-pty-evidence/blobs/c4286459b7175d84bcab8689a7609626e3ce0c3a022ef5a83020c986b514c14d/current.jsonl",
|
||||
"sha256": "110b0b2a183ecf81461313462b884d33b8ad303dc5f58c5366682f086a6d9fc3"
|
||||
},
|
||||
"records": [
|
||||
{
|
||||
"type": "assistant",
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-MPwj22",
|
||||
"sessionId": "0956daf4-4c61-4aa4-b163-2c429430386b",
|
||||
"isSidechain": false,
|
||||
"timestamp": "2026-09-09T18:24:32.177Z",
|
||||
"message": {
|
||||
"role": "assistant",
|
||||
"content": [
|
||||
{
|
||||
"type": "tool_use",
|
||||
"id": "toolu_01DhoRPXJYzVQXgXqCoXLfa4",
|
||||
"name": "AskUserQuestion",
|
||||
"input": {
|
||||
"questions": [
|
||||
{
|
||||
"header": "Routing",
|
||||
"question": "D1 \u2014 Add gstack skill routing rules to this repo's CLAUDE.md?\nProject/branch/task: gstack-plan-count fixture, branch main, reviewing PLAN.md (saved project views).\nELI10: gstack works best when CLAUDE.md tells the agent which slash skill to reach for (bugs \u2192 /investigate, ship \u2192 /ship, etc.). This is a one-time per-project setup. Plan mode is active, so if you say yes I will append and commit it only after the review finishes and plan mode exits.\nStakes if we pick wrong: Without routing, future sessions won't auto-route to the right skill; with it, one extra section lands in CLAUDE.md.\nRecommendation: A because it is a one-time, reversible convenience with no effect on this review.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nA) Add routing rules (recommended)\n \u2705 Future sessions pick the right gstack skill without you naming it each time\n \u2705 Reversible: delete the section from CLAUDE.md to undo, nothing else changes\n \u274c Adds ~15 lines to a fixture CLAUDE.md that currently only describes the review target\nB) No thanks, invoke skills manually\n \u2705 CLAUDE.md stays exactly as seeded, nothing touched outside the review\n \u2705 Recorded as declined so this prompt never repeats for this project\n \u274c You have to type skill names yourself in every future session here\nNet: convenience for future sessions vs. a pristine fixture file.",
|
||||
"options": [
|
||||
{
|
||||
"label": "Add routing rules (recommended)",
|
||||
"description": "Append the skill routing section to CLAUDE.md and commit it once plan mode exits."
|
||||
},
|
||||
{
|
||||
"label": "No thanks, manual",
|
||||
"description": "Record routing_declined=true; re-enable later with gstack-config set routing_declined false."
|
||||
}
|
||||
],
|
||||
"multiSelect": false
|
||||
},
|
||||
{
|
||||
"header": "Learnings",
|
||||
"question": "D2 \u2014 Enable cross-project learnings search?\nProject/branch/task: gstack-plan-count fixture, branch main, reviewing PLAN.md.\nELI10: gstack keeps a local file of lessons learned per project. It can also search lessons from your other projects on this machine to spot patterns that apply here. Nothing leaves your machine either way. Solo developers usually want this on; people juggling multiple client codebases usually keep it off.\nStakes if we pick wrong: Off means missing a relevant lesson from another repo; on means a note from a client repo could surface in an unrelated one.\nRecommendation: A because the search is local-only and the default for solo developers.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nA) Enable cross-project learnings (recommended)\n \u2705 Findings from your other repos on this machine inform this review and future ones\n \u2705 Stays local, no network, flip off any time with one config command\n \u274c Learnings from unrelated or client codebases can appear in this project's context\nB) Keep learnings project-scoped\n \u2705 Strict isolation between codebases, nothing from one repo shows up in another\n \u2705 Still records and reuses this project's own learnings normally\n \u274c Patterns you already learned elsewhere have to be rediscovered here\nNet: cross-pollination vs. strict per-repo isolation.",
|
||||
"options": [
|
||||
{
|
||||
"label": "Enable cross-project (recommended)",
|
||||
"description": "gstack-config set cross_project_learnings true, then search with --cross-project."
|
||||
},
|
||||
{
|
||||
"label": "Project-scoped only",
|
||||
"description": "gstack-config set cross_project_learnings false."
|
||||
}
|
||||
],
|
||||
"multiSelect": false
|
||||
},
|
||||
{
|
||||
"header": "Design doc",
|
||||
"question": "D3 \u2014 No design doc found for this branch. Run /office-hours first?\nProject/branch/task: gstack-plan-count fixture, branch main, reviewing PLAN.md.\nELI10: /office-hours produces a structured problem statement, a premise challenge, and explored alternatives before the review starts. It gives this CEO review sharper input, at the cost of roughly 10 minutes of Q&A first. The design doc is per-feature: it captures the thinking behind saved views specifically.\nStakes if we pick wrong: Skipping means the premise challenge in Step 0 leans on the 15-line PLAN.md alone; running it delays the review by ~10 minutes.\nRecommendation: B because PLAN.md already states the goal, approach, and validation clearly enough to challenge directly, and this is a standard review of a well-understood feature type.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nA) Run /office-hours now\n \u2705 Produces a written problem statement and alternatives the review can cite\n \u2705 Surfaces whether saved views is the right framing before we invest in review depth\n \u274c Adds ~10 minutes of interactive Q&A before any review finding appears\nB) Skip, proceed with standard review (recommended)\n \u2705 Review starts immediately; Step 0 premise challenge covers the same ground inline\n \u2705 PLAN.md already has a clear goal, approach, and validation to push against\n \u274c No persisted design doc for teammates to read alongside the plan\nNet: a written design artifact up front vs. getting to findings now.",
|
||||
"options": [
|
||||
{
|
||||
"label": "Run /office-hours now",
|
||||
"description": "Load and run the office-hours skill inline, then resume this review with its design doc."
|
||||
},
|
||||
{
|
||||
"label": "Skip (recommended)",
|
||||
"description": "Proceed straight to Step 0 premise challenge and approach alternatives."
|
||||
}
|
||||
],
|
||||
"multiSelect": false
|
||||
}
|
||||
]
|
||||
},
|
||||
"caller": {
|
||||
"type": "direct"
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
},
|
||||
{
|
||||
"type": "user",
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-MPwj22",
|
||||
"sessionId": "0956daf4-4c61-4aa4-b163-2c429430386b",
|
||||
"isSidechain": false,
|
||||
"timestamp": "2026-09-09T18:24:45.623Z",
|
||||
"message": {
|
||||
"role": "user",
|
||||
"content": [
|
||||
{
|
||||
"type": "tool_result",
|
||||
"content": "Your questions have been answered: \"D1 \u2014 Add gstack skill routing rules to this repo's CLAUDE.md?\nProject/branch/task: gstack-plan-count fixture, branch main, reviewing PLAN.md (saved project views).\nELI10: gstack works best when CLAUDE.md tells the agent which slash skill to reach for (bugs \u2192 /investigate, ship \u2192 /ship, etc.). This is a one-time per-project setup. Plan mode is active, so if you say yes I will append and commit it only after the review finishes and plan mode exits.\nStakes if we pick wrong: Without routing, future sessions won't auto-route to the right skill; with it, one extra section lands in CLAUDE.md.\nRecommendation: A because it is a one-time, reversible convenience with no effect on this review.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nA) Add routing rules (recommended)\n \u2705 Future sessions pick the right gstack skill without you naming it each time\n \u2705 Reversible: delete the section from CLAUDE.md to undo, nothing else changes\n \u274c Adds ~15 lines to a fixture CLAUDE.md that currently only describes the review target\nB) No thanks, invoke skills manually\n \u2705 CLAUDE.md stays exactly as seeded, nothing touched outside the review\n \u2705 Recorded as declined so this prompt never repeats for this project\n \u274c You have to type skill names yourself in every future session here\nNet: convenience for future sessions vs. a pristine fixture file.\"=\"Add routing rules (recommended)\", \"D2 \u2014 Enable cross-project learnings search?\nProject/branch/task: gstack-plan-count fixture, branch main, reviewing PLAN.md.\nELI10: gstack keeps a local file of lessons learned per project. It can also search lessons from your other projects on this machine to spot patterns that apply here. Nothing leaves your machine either way. Solo developers usually want this on; people juggling multiple client codebases usually keep it off.\nStakes if we pick wrong: Off means missing a relevant lesson from another repo; on means a note from a client repo could surface in an unrelated one.\nRecommendation: A because the search is local-only and the default for solo developers.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nA) Enable cross-project learnings (recommended)\n \u2705 Findings from your other repos on this machine inform this review and future ones\n \u2705 Stays local, no network, flip off any time with one config command\n \u274c Learnings from unrelated or client codebases can appear in this project's context\nB) Keep learnings project-scoped\n \u2705 Strict isolation between codebases, nothing from one repo shows up in another\n \u2705 Still records and reuses this project's own learnings normally\n \u274c Patterns you already learned elsewhere have to be rediscovered here\nNet: cross-pollination vs. strict per-repo isolation.\"=\"Enable cross-project (recommended)\", \"D3 \u2014 No design doc found for this branch. Run /office-hours first?\nProject/branch/task: gstack-plan-count fixture, branch main, reviewing PLAN.md.\nELI10: /office-hours produces a structured problem statement, a premise challenge, and explored alternatives before the review starts. It gives this CEO review sharper input, at the cost of roughly 10 minutes of Q&A first. The design doc is per-feature: it captures the thinking behind saved views specifically.\nStakes if we pick wrong: Skipping means the premise challenge in Step 0 leans on the 15-line PLAN.md alone; running it delays the review by ~10 minutes.\nRecommendation: B because PLAN.md already states the goal, approach, and validation clearly enough to challenge directly, and this is a standard review of a well-understood feature type.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nA) Run /office-hours now\n \u2705 Produces a written problem statement and alternatives the review can cite\n \u2705 Surfaces whether saved views is the right framing before we invest in review depth\n \u274c Adds ~10 minutes of interactive Q&A before any review finding appears\nB) Skip, proceed with standard review (recommended)\n \u2705 Review starts immediately; Step 0 premise challenge covers the same ground inline\n \u2705 PLAN.md already has a clear goal, approach, and validation to push against\n \u274c No persisted design doc for teammates to read alongside the plan\nNet: a written design artifact up front vs. getting to findings now.\"=\"Run /office-hours now\". You can now continue with these answers in mind.",
|
||||
"tool_use_id": "toolu_01DhoRPXJYzVQXgXqCoXLfa4"
|
||||
}
|
||||
]
|
||||
},
|
||||
"toolUseResult": {
|
||||
"questions": [
|
||||
{
|
||||
"question": "D1 \u2014 Add gstack skill routing rules to this repo's CLAUDE.md?\nProject/branch/task: gstack-plan-count fixture, branch main, reviewing PLAN.md (saved project views).\nELI10: gstack works best when CLAUDE.md tells the agent which slash skill to reach for (bugs \u2192 /investigate, ship \u2192 /ship, etc.). This is a one-time per-project setup. Plan mode is active, so if you say yes I will append and commit it only after the review finishes and plan mode exits.\nStakes if we pick wrong: Without routing, future sessions won't auto-route to the right skill; with it, one extra section lands in CLAUDE.md.\nRecommendation: A because it is a one-time, reversible convenience with no effect on this review.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nA) Add routing rules (recommended)\n \u2705 Future sessions pick the right gstack skill without you naming it each time\n \u2705 Reversible: delete the section from CLAUDE.md to undo, nothing else changes\n \u274c Adds ~15 lines to a fixture CLAUDE.md that currently only describes the review target\nB) No thanks, invoke skills manually\n \u2705 CLAUDE.md stays exactly as seeded, nothing touched outside the review\n \u2705 Recorded as declined so this prompt never repeats for this project\n \u274c You have to type skill names yourself in every future session here\nNet: convenience for future sessions vs. a pristine fixture file.",
|
||||
"header": "Routing",
|
||||
"options": [
|
||||
{
|
||||
"label": "Add routing rules (recommended)",
|
||||
"description": "Append the skill routing section to CLAUDE.md and commit it once plan mode exits."
|
||||
},
|
||||
{
|
||||
"label": "No thanks, manual",
|
||||
"description": "Record routing_declined=true; re-enable later with gstack-config set routing_declined false."
|
||||
}
|
||||
],
|
||||
"multiSelect": false
|
||||
},
|
||||
{
|
||||
"question": "D2 \u2014 Enable cross-project learnings search?\nProject/branch/task: gstack-plan-count fixture, branch main, reviewing PLAN.md.\nELI10: gstack keeps a local file of lessons learned per project. It can also search lessons from your other projects on this machine to spot patterns that apply here. Nothing leaves your machine either way. Solo developers usually want this on; people juggling multiple client codebases usually keep it off.\nStakes if we pick wrong: Off means missing a relevant lesson from another repo; on means a note from a client repo could surface in an unrelated one.\nRecommendation: A because the search is local-only and the default for solo developers.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nA) Enable cross-project learnings (recommended)\n \u2705 Findings from your other repos on this machine inform this review and future ones\n \u2705 Stays local, no network, flip off any time with one config command\n \u274c Learnings from unrelated or client codebases can appear in this project's context\nB) Keep learnings project-scoped\n \u2705 Strict isolation between codebases, nothing from one repo shows up in another\n \u2705 Still records and reuses this project's own learnings normally\n \u274c Patterns you already learned elsewhere have to be rediscovered here\nNet: cross-pollination vs. strict per-repo isolation.",
|
||||
"header": "Learnings",
|
||||
"options": [
|
||||
{
|
||||
"label": "Enable cross-project (recommended)",
|
||||
"description": "gstack-config set cross_project_learnings true, then search with --cross-project."
|
||||
},
|
||||
{
|
||||
"label": "Project-scoped only",
|
||||
"description": "gstack-config set cross_project_learnings false."
|
||||
}
|
||||
],
|
||||
"multiSelect": false
|
||||
},
|
||||
{
|
||||
"question": "D3 \u2014 No design doc found for this branch. Run /office-hours first?\nProject/branch/task: gstack-plan-count fixture, branch main, reviewing PLAN.md.\nELI10: /office-hours produces a structured problem statement, a premise challenge, and explored alternatives before the review starts. It gives this CEO review sharper input, at the cost of roughly 10 minutes of Q&A first. The design doc is per-feature: it captures the thinking behind saved views specifically.\nStakes if we pick wrong: Skipping means the premise challenge in Step 0 leans on the 15-line PLAN.md alone; running it delays the review by ~10 minutes.\nRecommendation: B because PLAN.md already states the goal, approach, and validation clearly enough to challenge directly, and this is a standard review of a well-understood feature type.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nA) Run /office-hours now\n \u2705 Produces a written problem statement and alternatives the review can cite\n \u2705 Surfaces whether saved views is the right framing before we invest in review depth\n \u274c Adds ~10 minutes of interactive Q&A before any review finding appears\nB) Skip, proceed with standard review (recommended)\n \u2705 Review starts immediately; Step 0 premise challenge covers the same ground inline\n \u2705 PLAN.md already has a clear goal, approach, and validation to push against\n \u274c No persisted design doc for teammates to read alongside the plan\nNet: a written design artifact up front vs. getting to findings now.",
|
||||
"header": "Design doc",
|
||||
"options": [
|
||||
{
|
||||
"label": "Run /office-hours now",
|
||||
"description": "Load and run the office-hours skill inline, then resume this review with its design doc."
|
||||
},
|
||||
{
|
||||
"label": "Skip (recommended)",
|
||||
"description": "Proceed straight to Step 0 premise challenge and approach alternatives."
|
||||
}
|
||||
],
|
||||
"multiSelect": false
|
||||
}
|
||||
],
|
||||
"answers": {
|
||||
"D1 \u2014 Add gstack skill routing rules to this repo's CLAUDE.md?\nProject/branch/task: gstack-plan-count fixture, branch main, reviewing PLAN.md (saved project views).\nELI10: gstack works best when CLAUDE.md tells the agent which slash skill to reach for (bugs \u2192 /investigate, ship \u2192 /ship, etc.). This is a one-time per-project setup. Plan mode is active, so if you say yes I will append and commit it only after the review finishes and plan mode exits.\nStakes if we pick wrong: Without routing, future sessions won't auto-route to the right skill; with it, one extra section lands in CLAUDE.md.\nRecommendation: A because it is a one-time, reversible convenience with no effect on this review.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nA) Add routing rules (recommended)\n \u2705 Future sessions pick the right gstack skill without you naming it each time\n \u2705 Reversible: delete the section from CLAUDE.md to undo, nothing else changes\n \u274c Adds ~15 lines to a fixture CLAUDE.md that currently only describes the review target\nB) No thanks, invoke skills manually\n \u2705 CLAUDE.md stays exactly as seeded, nothing touched outside the review\n \u2705 Recorded as declined so this prompt never repeats for this project\n \u274c You have to type skill names yourself in every future session here\nNet: convenience for future sessions vs. a pristine fixture file.": "Add routing rules (recommended)",
|
||||
"D2 \u2014 Enable cross-project learnings search?\nProject/branch/task: gstack-plan-count fixture, branch main, reviewing PLAN.md.\nELI10: gstack keeps a local file of lessons learned per project. It can also search lessons from your other projects on this machine to spot patterns that apply here. Nothing leaves your machine either way. Solo developers usually want this on; people juggling multiple client codebases usually keep it off.\nStakes if we pick wrong: Off means missing a relevant lesson from another repo; on means a note from a client repo could surface in an unrelated one.\nRecommendation: A because the search is local-only and the default for solo developers.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nA) Enable cross-project learnings (recommended)\n \u2705 Findings from your other repos on this machine inform this review and future ones\n \u2705 Stays local, no network, flip off any time with one config command\n \u274c Learnings from unrelated or client codebases can appear in this project's context\nB) Keep learnings project-scoped\n \u2705 Strict isolation between codebases, nothing from one repo shows up in another\n \u2705 Still records and reuses this project's own learnings normally\n \u274c Patterns you already learned elsewhere have to be rediscovered here\nNet: cross-pollination vs. strict per-repo isolation.": "Enable cross-project (recommended)",
|
||||
"D3 \u2014 No design doc found for this branch. Run /office-hours first?\nProject/branch/task: gstack-plan-count fixture, branch main, reviewing PLAN.md.\nELI10: /office-hours produces a structured problem statement, a premise challenge, and explored alternatives before the review starts. It gives this CEO review sharper input, at the cost of roughly 10 minutes of Q&A first. The design doc is per-feature: it captures the thinking behind saved views specifically.\nStakes if we pick wrong: Skipping means the premise challenge in Step 0 leans on the 15-line PLAN.md alone; running it delays the review by ~10 minutes.\nRecommendation: B because PLAN.md already states the goal, approach, and validation clearly enough to challenge directly, and this is a standard review of a well-understood feature type.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nA) Run /office-hours now\n \u2705 Produces a written problem statement and alternatives the review can cite\n \u2705 Surfaces whether saved views is the right framing before we invest in review depth\n \u274c Adds ~10 minutes of interactive Q&A before any review finding appears\nB) Skip, proceed with standard review (recommended)\n \u2705 Review starts immediately; Step 0 premise challenge covers the same ground inline\n \u2705 PLAN.md already has a clear goal, approach, and validation to push against\n \u274c No persisted design doc for teammates to read alongside the plan\nNet: a written design artifact up front vs. getting to findings now.": "Run /office-hours now"
|
||||
},
|
||||
"annotations": {}
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"targetMode": "SCOPE EXPANSION",
|
||||
"attempt": "plan-ceo-review-mode-1788978153191-KJ7XbV",
|
||||
"actualState": "failed",
|
||||
"selectedAt": 1788978337334,
|
||||
"commandStartedAt": 1788978151180,
|
||||
"pendingQuestionRecorder": {
|
||||
"status": "pending"
|
||||
},
|
||||
"process": {
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-6qKEas",
|
||||
"sessionId": "20ece027-6cbd-4681-b96e-7c4eeb9c0e3c"
|
||||
},
|
||||
"observation": {
|
||||
"path": ".context/ship-source-ad-full-paid-20260909-v1/evals/job-7/shards/skill-e2e-plan-ceo-mode-routing/pty-count/ship-source-ad-full-paid-20260909-v1-7/plan-ceo-review-mode-1788978153191-KJ7XbV/observation.json",
|
||||
"sha256": "8bfd5f8b39ed5ac713c942b90f8fe32df9c3d69b44a422fa633d51dc7dba2c04"
|
||||
},
|
||||
"native": {
|
||||
"captureKey": "native:1349451-870044:20ece027-6cbd-4681-b96e-7c4eeb9c0e3c:54172659",
|
||||
"path": "/home/vercel-sandbox/gstack/.context/ship-source-ad-full-paid-20260909-v1/full-pty-evidence/blobs/c24085222d7bed850114940c3d9a281821ec87df662bf200dd676d1c345b3a29/current.jsonl",
|
||||
"sha256": "4dd60d3c5622524e5427164101f57cce31a04ae31f6fc1c195aeae9445082855"
|
||||
},
|
||||
"records": [
|
||||
{
|
||||
"type": "assistant",
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-6qKEas",
|
||||
"sessionId": "20ece027-6cbd-4681-b96e-7c4eeb9c0e3c",
|
||||
"isSidechain": false,
|
||||
"timestamp": "2026-09-09T18:25:35.909Z",
|
||||
"message": {
|
||||
"role": "assistant",
|
||||
"content": [
|
||||
{
|
||||
"type": "tool_use",
|
||||
"id": "toolu_01GFkLSt2i2McBknF87PonJh",
|
||||
"name": "AskUserQuestion",
|
||||
"input": {
|
||||
"questions": [
|
||||
{
|
||||
"question": "D5 \u2014 Which review posture for the rest of this CEO review?\nProject/branch/task: main, PLAN.md \"Add saved project views\", approach B (spec-based views) selected.\nELI10: The mode sets how ambitious the rest of the review is. Expansion pushes scope up and recommends extras enthusiastically. Selective expansion keeps your scope as the baseline but shows each extra as its own yes/no so you cherry-pick. Hold scope only hardens what is already there. Reduction strips to the bare minimum. Whatever you pick, no scope changes without your explicit yes on each one.\nStakes if we pick wrong: Too ambitious and a small productivity feature grows into a platform project; too narrow and the obvious adjacent wins (default view, shareable link) get rebuilt later at higher cost.\nRecommendation: SELECTIVE EXPANSION because this enhances an existing task list rather than building a greenfield feature, and approach B already opens a few cheap adjacent doors worth a look.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nNet: how many adjacent opportunities you want to see before the rigor sections start.",
|
||||
"header": "Mode",
|
||||
"multiSelect": false,
|
||||
"options": [
|
||||
{
|
||||
"label": "SELECTIVE EXPANSION (recommended)",
|
||||
"description": "\u2705 Plan scope stays the baseline and gets the full rigor pass\n\u2705 Each expansion (default view, shareable link, shared views, etc.) is a separate neutral yes/no you control\n\u274c Adds several decision prompts before the deep review begins"
|
||||
},
|
||||
{
|
||||
"label": "SCOPE EXPANSION",
|
||||
"description": "\u2705 Surfaces the 10x version and platonic ideal, recommended enthusiastically\n\u2705 Best if you want saved views to become a real product pillar\n\u274c Risks turning a two-day feature into a multi-week views platform"
|
||||
},
|
||||
{
|
||||
"label": "HOLD SCOPE",
|
||||
"description": "\u2705 Maximum rigor on exactly the four approach bullets, nothing else surfaced\n\u2705 Fastest path to a bulletproof plan for the stated goal\n\u274c Cheap adjacent wins go unmentioned and may be rebuilt later"
|
||||
},
|
||||
{
|
||||
"label": "SCOPE REDUCTION",
|
||||
"description": "\u2705 Finds the smallest shippable slice, useful if the pilot is really a demand test\n\u2705 Minimizes new attack surface and migration risk\n\u274c Approach B already leans complete; cutting now would undo that choice"
|
||||
}
|
||||
]
|
||||
}
|
||||
]
|
||||
},
|
||||
"caller": {
|
||||
"type": "direct"
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
},
|
||||
{
|
||||
"type": "user",
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-6qKEas",
|
||||
"sessionId": "20ece027-6cbd-4681-b96e-7c4eeb9c0e3c",
|
||||
"isSidechain": false,
|
||||
"timestamp": "2026-09-09T18:25:37.369Z",
|
||||
"message": {
|
||||
"role": "user",
|
||||
"content": [
|
||||
{
|
||||
"type": "tool_result",
|
||||
"content": "Your questions have been answered: \"D5 \u2014 Which review posture for the rest of this CEO review?\nProject/branch/task: main, PLAN.md \"Add saved project views\", approach B (spec-based views) selected.\nELI10: The mode sets how ambitious the rest of the review is. Expansion pushes scope up and recommends extras enthusiastically. Selective expansion keeps your scope as the baseline but shows each extra as its own yes/no so you cherry-pick. Hold scope only hardens what is already there. Reduction strips to the bare minimum. Whatever you pick, no scope changes without your explicit yes on each one.\nStakes if we pick wrong: Too ambitious and a small productivity feature grows into a platform project; too narrow and the obvious adjacent wins (default view, shareable link) get rebuilt later at higher cost.\nRecommendation: SELECTIVE EXPANSION because this enhances an existing task list rather than building a greenfield feature, and approach B already opens a few cheap adjacent doors worth a look.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nNet: how many adjacent opportunities you want to see before the rigor sections start.\"=\"SCOPE EXPANSION\". You can now continue with these answers in mind.",
|
||||
"tool_use_id": "toolu_01GFkLSt2i2McBknF87PonJh"
|
||||
}
|
||||
]
|
||||
},
|
||||
"toolUseResult": {
|
||||
"questions": [
|
||||
{
|
||||
"question": "D5 \u2014 Which review posture for the rest of this CEO review?\nProject/branch/task: main, PLAN.md \"Add saved project views\", approach B (spec-based views) selected.\nELI10: The mode sets how ambitious the rest of the review is. Expansion pushes scope up and recommends extras enthusiastically. Selective expansion keeps your scope as the baseline but shows each extra as its own yes/no so you cherry-pick. Hold scope only hardens what is already there. Reduction strips to the bare minimum. Whatever you pick, no scope changes without your explicit yes on each one.\nStakes if we pick wrong: Too ambitious and a small productivity feature grows into a platform project; too narrow and the obvious adjacent wins (default view, shareable link) get rebuilt later at higher cost.\nRecommendation: SELECTIVE EXPANSION because this enhances an existing task list rather than building a greenfield feature, and approach B already opens a few cheap adjacent doors worth a look.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nNet: how many adjacent opportunities you want to see before the rigor sections start.",
|
||||
"header": "Mode",
|
||||
"options": [
|
||||
{
|
||||
"label": "SELECTIVE EXPANSION (recommended)",
|
||||
"description": "\u2705 Plan scope stays the baseline and gets the full rigor pass\n\u2705 Each expansion (default view, shareable link, shared views, etc.) is a separate neutral yes/no you control\n\u274c Adds several decision prompts before the deep review begins"
|
||||
},
|
||||
{
|
||||
"label": "SCOPE EXPANSION",
|
||||
"description": "\u2705 Surfaces the 10x version and platonic ideal, recommended enthusiastically\n\u2705 Best if you want saved views to become a real product pillar\n\u274c Risks turning a two-day feature into a multi-week views platform"
|
||||
},
|
||||
{
|
||||
"label": "HOLD SCOPE",
|
||||
"description": "\u2705 Maximum rigor on exactly the four approach bullets, nothing else surfaced\n\u2705 Fastest path to a bulletproof plan for the stated goal\n\u274c Cheap adjacent wins go unmentioned and may be rebuilt later"
|
||||
},
|
||||
{
|
||||
"label": "SCOPE REDUCTION",
|
||||
"description": "\u2705 Finds the smallest shippable slice, useful if the pilot is really a demand test\n\u2705 Minimizes new attack surface and migration risk\n\u274c Approach B already leans complete; cutting now would undo that choice"
|
||||
}
|
||||
],
|
||||
"multiSelect": false
|
||||
}
|
||||
],
|
||||
"answers": {
|
||||
"D5 \u2014 Which review posture for the rest of this CEO review?\nProject/branch/task: main, PLAN.md \"Add saved project views\", approach B (spec-based views) selected.\nELI10: The mode sets how ambitious the rest of the review is. Expansion pushes scope up and recommends extras enthusiastically. Selective expansion keeps your scope as the baseline but shows each extra as its own yes/no so you cherry-pick. Hold scope only hardens what is already there. Reduction strips to the bare minimum. Whatever you pick, no scope changes without your explicit yes on each one.\nStakes if we pick wrong: Too ambitious and a small productivity feature grows into a platform project; too narrow and the obvious adjacent wins (default view, shareable link) get rebuilt later at higher cost.\nRecommendation: SELECTIVE EXPANSION because this enhances an existing task list rather than building a greenfield feature, and approach B already opens a few cheap adjacent doors worth a look.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nNet: how many adjacent opportunities you want to see before the rigor sections start.": "SCOPE EXPANSION"
|
||||
},
|
||||
"annotations": {}
|
||||
}
|
||||
},
|
||||
{
|
||||
"type": "assistant",
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-6qKEas",
|
||||
"sessionId": "20ece027-6cbd-4681-b96e-7c4eeb9c0e3c",
|
||||
"isSidechain": false,
|
||||
"timestamp": "2026-09-09T18:26:17.497Z",
|
||||
"message": {
|
||||
"role": "assistant",
|
||||
"content": [
|
||||
{
|
||||
"type": "tool_use",
|
||||
"id": "toolu_01VUTgcQria2X3TVg941K3aV",
|
||||
"name": "AskUserQuestion",
|
||||
"input": {
|
||||
"questions": [
|
||||
{
|
||||
"question": "D6 \u2014 Proposal 1: Default view per member per project?\nProject/branch/task: main, PLAN.md saved views, SCOPE EXPANSION opt-in ceremony (1 of 6).\nELI10: Imagine opening a project and the task list is already filtered the way you left it, no clicking required. Each member can mark one saved view as their default for a project; the list opens in that view, with a visible chip showing which view is active and a one-click way back to \"All tasks\". Concrete shape: a default_view_id on the project membership (or an is_default flag on saved_views with a unique partial index), read at task-list load. Effort: human ~half day / CC ~10 min.\nStakes if we pick wrong: Without it, members still start from the unfiltered list every session and reuse depends on remembering the picker exists, which is exactly how saved-filter features die.\nRecommendation: Add because it turns saved views from a thing you remember to use into the thing you land on, which is the single biggest lever on the plan's reuse metric.\nCompleteness: A=10/10, B=5/10, C=0/10\nNet: one flag and one load-time lookup buy the habit loop the pilot needs.",
|
||||
"header": "Proposal 1",
|
||||
"multiSelect": false,
|
||||
"options": [
|
||||
{
|
||||
"label": "A) Add to this plan's scope (recommended)",
|
||||
"description": "\u2705 Members land in their view automatically; reuse becomes passive instead of remembered\n\u2705 Trivial schema: partial unique index on (member_id, project_id) where is_default\n\u274c Needs a clear \"you're in a filtered view\" indicator so newcomers aren't confused by missing tasks"
|
||||
},
|
||||
{
|
||||
"label": "B) Defer to TODOS.md",
|
||||
"description": "\u2705 Keeps the first ship focused on save and reopen only\n\u2705 Can be added later without a data migration if is_default is nullable\n\u274c Pilot measures reuse without the mechanism most likely to drive it"
|
||||
},
|
||||
{
|
||||
"label": "C) Skip",
|
||||
"description": "\u2705 Smallest possible surface; no default-state confusion to design for\n\u2705 Zero additional edge cases (deleted default view, etc.)\n\u274c Leaves the habit loop entirely to the user's memory"
|
||||
}
|
||||
]
|
||||
}
|
||||
]
|
||||
},
|
||||
"caller": {
|
||||
"type": "direct"
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
},
|
||||
{
|
||||
"type": "user",
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-6qKEas",
|
||||
"sessionId": "20ece027-6cbd-4681-b96e-7c4eeb9c0e3c",
|
||||
"isSidechain": false,
|
||||
"timestamp": "2026-09-09T18:26:20.110Z",
|
||||
"message": {
|
||||
"role": "user",
|
||||
"content": [
|
||||
{
|
||||
"type": "tool_result",
|
||||
"content": "Your questions have been answered: \"D6 \u2014 Proposal 1: Default view per member per project?\nProject/branch/task: main, PLAN.md saved views, SCOPE EXPANSION opt-in ceremony (1 of 6).\nELI10: Imagine opening a project and the task list is already filtered the way you left it, no clicking required. Each member can mark one saved view as their default for a project; the list opens in that view, with a visible chip showing which view is active and a one-click way back to \"All tasks\". Concrete shape: a default_view_id on the project membership (or an is_default flag on saved_views with a unique partial index), read at task-list load. Effort: human ~half day / CC ~10 min.\nStakes if we pick wrong: Without it, members still start from the unfiltered list every session and reuse depends on remembering the picker exists, which is exactly how saved-filter features die.\nRecommendation: Add because it turns saved views from a thing you remember to use into the thing you land on, which is the single biggest lever on the plan's reuse metric.\nCompleteness: A=10/10, B=5/10, C=0/10\nNet: one flag and one load-time lookup buy the habit loop the pilot needs.\"=\"A) Add to this plan's scope (recommended)\". You can now continue with these answers in mind.",
|
||||
"tool_use_id": "toolu_01VUTgcQria2X3TVg941K3aV"
|
||||
}
|
||||
]
|
||||
},
|
||||
"toolUseResult": {
|
||||
"questions": [
|
||||
{
|
||||
"question": "D6 \u2014 Proposal 1: Default view per member per project?\nProject/branch/task: main, PLAN.md saved views, SCOPE EXPANSION opt-in ceremony (1 of 6).\nELI10: Imagine opening a project and the task list is already filtered the way you left it, no clicking required. Each member can mark one saved view as their default for a project; the list opens in that view, with a visible chip showing which view is active and a one-click way back to \"All tasks\". Concrete shape: a default_view_id on the project membership (or an is_default flag on saved_views with a unique partial index), read at task-list load. Effort: human ~half day / CC ~10 min.\nStakes if we pick wrong: Without it, members still start from the unfiltered list every session and reuse depends on remembering the picker exists, which is exactly how saved-filter features die.\nRecommendation: Add because it turns saved views from a thing you remember to use into the thing you land on, which is the single biggest lever on the plan's reuse metric.\nCompleteness: A=10/10, B=5/10, C=0/10\nNet: one flag and one load-time lookup buy the habit loop the pilot needs.",
|
||||
"header": "Proposal 1",
|
||||
"options": [
|
||||
{
|
||||
"label": "A) Add to this plan's scope (recommended)",
|
||||
"description": "\u2705 Members land in their view automatically; reuse becomes passive instead of remembered\n\u2705 Trivial schema: partial unique index on (member_id, project_id) where is_default\n\u274c Needs a clear \"you're in a filtered view\" indicator so newcomers aren't confused by missing tasks"
|
||||
},
|
||||
{
|
||||
"label": "B) Defer to TODOS.md",
|
||||
"description": "\u2705 Keeps the first ship focused on save and reopen only\n\u2705 Can be added later without a data migration if is_default is nullable\n\u274c Pilot measures reuse without the mechanism most likely to drive it"
|
||||
},
|
||||
{
|
||||
"label": "C) Skip",
|
||||
"description": "\u2705 Smallest possible surface; no default-state confusion to design for\n\u2705 Zero additional edge cases (deleted default view, etc.)\n\u274c Leaves the habit loop entirely to the user's memory"
|
||||
}
|
||||
],
|
||||
"multiSelect": false
|
||||
}
|
||||
],
|
||||
"answers": {
|
||||
"D6 \u2014 Proposal 1: Default view per member per project?\nProject/branch/task: main, PLAN.md saved views, SCOPE EXPANSION opt-in ceremony (1 of 6).\nELI10: Imagine opening a project and the task list is already filtered the way you left it, no clicking required. Each member can mark one saved view as their default for a project; the list opens in that view, with a visible chip showing which view is active and a one-click way back to \"All tasks\". Concrete shape: a default_view_id on the project membership (or an is_default flag on saved_views with a unique partial index), read at task-list load. Effort: human ~half day / CC ~10 min.\nStakes if we pick wrong: Without it, members still start from the unfiltered list every session and reuse depends on remembering the picker exists, which is exactly how saved-filter features die.\nRecommendation: Add because it turns saved views from a thing you remember to use into the thing you land on, which is the single biggest lever on the plan's reuse metric.\nCompleteness: A=10/10, B=5/10, C=0/10\nNet: one flag and one load-time lookup buy the habit loop the pilot needs.": "A) Add to this plan's scope (recommended)"
|
||||
},
|
||||
"annotations": {}
|
||||
}
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"targetMode": "HOLD SCOPE",
|
||||
"attempt": "plan-ceo-review-mode-1788978583199-DTJLos",
|
||||
"actualState": "failed",
|
||||
"selectedAt": 1788978796772,
|
||||
"commandStartedAt": 1788978581188,
|
||||
"pendingQuestionRecorder": {
|
||||
"status": "pending"
|
||||
},
|
||||
"process": {
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-H67mSL",
|
||||
"sessionId": "af7bf1c9-f26a-4b9b-96cc-02005d094575"
|
||||
},
|
||||
"observation": {
|
||||
"path": ".context/ship-source-ad-full-paid-20260909-v1/evals/job-7/shards/skill-e2e-plan-ceo-mode-routing/pty-count/ship-source-ad-full-paid-20260909-v1-7/plan-ceo-review-mode-1788978583199-DTJLos/observation.json",
|
||||
"sha256": "290fa1399bc1450f3be43f91fdc5adf2f69b93e26e47360193c3e5545830a843"
|
||||
},
|
||||
"native": {
|
||||
"captureKey": "native:1370304-913045:af7bf1c9-f26a-4b9b-96cc-02005d094575:53667299",
|
||||
"path": "/home/vercel-sandbox/gstack/.context/ship-source-ad-full-paid-20260909-v1/full-pty-evidence/blobs/4252e3332946e781b0d402c72f21ac5bbfd431469f156122f02f3531ebab7026/current.jsonl",
|
||||
"sha256": "6367bf198c039887017ddb6f24da5d0d207d10dc2d90e28e2a3376d787899fdd"
|
||||
},
|
||||
"records": [
|
||||
{
|
||||
"type": "assistant",
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-H67mSL",
|
||||
"sessionId": "af7bf1c9-f26a-4b9b-96cc-02005d094575",
|
||||
"isSidechain": false,
|
||||
"timestamp": "2026-09-09T18:32:48.066Z",
|
||||
"message": {
|
||||
"role": "assistant",
|
||||
"content": [
|
||||
{
|
||||
"type": "tool_use",
|
||||
"id": "toolu_01GPP2rDM3uB3dwT54UhvG7i",
|
||||
"name": "AskUserQuestion",
|
||||
"input": {
|
||||
"questions": [
|
||||
{
|
||||
"header": "Approach",
|
||||
"question": "D3 \u2014 Which implementation approach should the plan commit to?\nProject/branch/task: gstack-plan-count-H67mSL on main, PLAN.md saved project views.\nELI10: All three approaches let a member stop re-building filters, but they differ in how much of the future they leave open. A saves a blob of JSON per view and moves on. B first defines one shape for \"what the list looks like right now\" and uses that same shape for the URL, the saved view, and the remembered last filter. C skips naming entirely and just makes the URL and last-used memory do the work. The stakes: with an opaque blob, renaming a filter field later silently breaks every saved view pointing at it, and adding shared team views means a migration.\nStakes if we pick wrong: A ships fast but the first filter rename produces views that open blank with no error; C never meets the goal of named views.\nRecommendation: B because one serialization is the DRY answer, stale views degrade visibly instead of silently, and sharing later is a column flip. The extra work is roughly 25 minutes of CC time.\nCompleteness: A=6/10, B=9/10, C=4/10\nNet: about 25 extra minutes of CC time buys a schema that the URL, memory, and sharing can all reuse instead of three parallel state shapes.",
|
||||
"options": [
|
||||
{
|
||||
"label": "B: ViewState schema (recommended)",
|
||||
"description": "Completeness 9/10. Effort M (human ~4 days / CC ~45 min).\n\u2705 One versioned ViewState shape drives URL params, saved views, and per-member last-used memory\n\u2705 Loading a view validates against current filter vocabulary and shows a notice instead of a blank list\n\u2705 visibility, is_default, and position columns mean sharing and defaults need no migration later\n\u274c More upfront schema design and a versioning story for the state column"
|
||||
},
|
||||
{
|
||||
"label": "A: Minimal table",
|
||||
"description": "Completeness 6/10. Effort S (human ~2 days / CC ~20 min).\n\u2705 Smallest diff that ships exactly what PLAN.md describes today\n\u2705 Easy to reason about: one table, four endpoints, one dropdown\n\u274c Opaque JSONB means a renamed filter field breaks saved views silently with no user-facing error"
|
||||
},
|
||||
{
|
||||
"label": "C: URL + last-used only",
|
||||
"description": "Completeness 4/10. Effort S (human ~1 day / CC ~15 min).\n\u2705 No new table or endpoints, filters survive reload and are shareable by link immediately\n\u2705 Lowest risk surface, nothing new to secure or migrate\n\u274c Does not deliver named, reopenable views, which is the stated goal of the plan"
|
||||
}
|
||||
]
|
||||
}
|
||||
]
|
||||
},
|
||||
"caller": {
|
||||
"type": "direct"
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
},
|
||||
{
|
||||
"type": "user",
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-H67mSL",
|
||||
"sessionId": "af7bf1c9-f26a-4b9b-96cc-02005d094575",
|
||||
"isSidechain": false,
|
||||
"timestamp": "2026-09-09T18:32:48.565Z",
|
||||
"message": {
|
||||
"role": "user",
|
||||
"content": [
|
||||
{
|
||||
"type": "tool_result",
|
||||
"content": "Your questions have been answered: \"D3 \u2014 Which implementation approach should the plan commit to?\nProject/branch/task: gstack-plan-count-H67mSL on main, PLAN.md saved project views.\nELI10: All three approaches let a member stop re-building filters, but they differ in how much of the future they leave open. A saves a blob of JSON per view and moves on. B first defines one shape for \"what the list looks like right now\" and uses that same shape for the URL, the saved view, and the remembered last filter. C skips naming entirely and just makes the URL and last-used memory do the work. The stakes: with an opaque blob, renaming a filter field later silently breaks every saved view pointing at it, and adding shared team views means a migration.\nStakes if we pick wrong: A ships fast but the first filter rename produces views that open blank with no error; C never meets the goal of named views.\nRecommendation: B because one serialization is the DRY answer, stale views degrade visibly instead of silently, and sharing later is a column flip. The extra work is roughly 25 minutes of CC time.\nCompleteness: A=6/10, B=9/10, C=4/10\nNet: about 25 extra minutes of CC time buys a schema that the URL, memory, and sharing can all reuse instead of three parallel state shapes.\"=\"B: ViewState schema (recommended)\". You can now continue with these answers in mind.",
|
||||
"tool_use_id": "toolu_01GPP2rDM3uB3dwT54UhvG7i"
|
||||
}
|
||||
]
|
||||
},
|
||||
"toolUseResult": {
|
||||
"questions": [
|
||||
{
|
||||
"question": "D3 \u2014 Which implementation approach should the plan commit to?\nProject/branch/task: gstack-plan-count-H67mSL on main, PLAN.md saved project views.\nELI10: All three approaches let a member stop re-building filters, but they differ in how much of the future they leave open. A saves a blob of JSON per view and moves on. B first defines one shape for \"what the list looks like right now\" and uses that same shape for the URL, the saved view, and the remembered last filter. C skips naming entirely and just makes the URL and last-used memory do the work. The stakes: with an opaque blob, renaming a filter field later silently breaks every saved view pointing at it, and adding shared team views means a migration.\nStakes if we pick wrong: A ships fast but the first filter rename produces views that open blank with no error; C never meets the goal of named views.\nRecommendation: B because one serialization is the DRY answer, stale views degrade visibly instead of silently, and sharing later is a column flip. The extra work is roughly 25 minutes of CC time.\nCompleteness: A=6/10, B=9/10, C=4/10\nNet: about 25 extra minutes of CC time buys a schema that the URL, memory, and sharing can all reuse instead of three parallel state shapes.",
|
||||
"header": "Approach",
|
||||
"options": [
|
||||
{
|
||||
"label": "B: ViewState schema (recommended)",
|
||||
"description": "Completeness 9/10. Effort M (human ~4 days / CC ~45 min).\n\u2705 One versioned ViewState shape drives URL params, saved views, and per-member last-used memory\n\u2705 Loading a view validates against current filter vocabulary and shows a notice instead of a blank list\n\u2705 visibility, is_default, and position columns mean sharing and defaults need no migration later\n\u274c More upfront schema design and a versioning story for the state column"
|
||||
},
|
||||
{
|
||||
"label": "A: Minimal table",
|
||||
"description": "Completeness 6/10. Effort S (human ~2 days / CC ~20 min).\n\u2705 Smallest diff that ships exactly what PLAN.md describes today\n\u2705 Easy to reason about: one table, four endpoints, one dropdown\n\u274c Opaque JSONB means a renamed filter field breaks saved views silently with no user-facing error"
|
||||
},
|
||||
{
|
||||
"label": "C: URL + last-used only",
|
||||
"description": "Completeness 4/10. Effort S (human ~1 day / CC ~15 min).\n\u2705 No new table or endpoints, filters survive reload and are shareable by link immediately\n\u2705 Lowest risk surface, nothing new to secure or migrate\n\u274c Does not deliver named, reopenable views, which is the stated goal of the plan"
|
||||
}
|
||||
],
|
||||
"multiSelect": false
|
||||
}
|
||||
],
|
||||
"answers": {
|
||||
"D3 \u2014 Which implementation approach should the plan commit to?\nProject/branch/task: gstack-plan-count-H67mSL on main, PLAN.md saved project views.\nELI10: All three approaches let a member stop re-building filters, but they differ in how much of the future they leave open. A saves a blob of JSON per view and moves on. B first defines one shape for \"what the list looks like right now\" and uses that same shape for the URL, the saved view, and the remembered last filter. C skips naming entirely and just makes the URL and last-used memory do the work. The stakes: with an opaque blob, renaming a filter field later silently breaks every saved view pointing at it, and adding shared team views means a migration.\nStakes if we pick wrong: A ships fast but the first filter rename produces views that open blank with no error; C never meets the goal of named views.\nRecommendation: B because one serialization is the DRY answer, stale views degrade visibly instead of silently, and sharing later is a column flip. The extra work is roughly 25 minutes of CC time.\nCompleteness: A=6/10, B=9/10, C=4/10\nNet: about 25 extra minutes of CC time buys a schema that the URL, memory, and sharing can all reuse instead of three parallel state shapes.": "B: ViewState schema (recommended)"
|
||||
},
|
||||
"annotations": {}
|
||||
}
|
||||
},
|
||||
{
|
||||
"type": "assistant",
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-H67mSL",
|
||||
"sessionId": "af7bf1c9-f26a-4b9b-96cc-02005d094575",
|
||||
"isSidechain": false,
|
||||
"timestamp": "2026-09-09T18:33:16.411Z",
|
||||
"message": {
|
||||
"role": "assistant",
|
||||
"content": [
|
||||
{
|
||||
"type": "tool_use",
|
||||
"id": "toolu_01LfRN9N3Fi92y3iQCREW1xT",
|
||||
"name": "AskUserQuestion",
|
||||
"input": {
|
||||
"questions": [
|
||||
{
|
||||
"header": "Review mode",
|
||||
"question": "D4 \u2014 Which review posture should I take for the rest of this review?\nProject/branch/task: gstack-plan-count-H67mSL on main, PLAN.md saved project views, approach B chosen.\nELI10: The approach is settled. Now I need to know how ambitious to be. Expansion means I dream up the cathedral version and pitch each piece to you one at a time. Selective expansion means I hold your scope as the baseline, make it bulletproof, and separately show you each add-on so you can cherry-pick with no sales pitch. Hold scope means no add-ons at all, just maximum rigor on what is written. Reduction means I cut the plan to the smallest thing that stops members re-building filters. Every scope change in every mode is still your explicit opt-in.\nStakes if we pick wrong: too ambitious and the pilot slips; too narrow and you ship a table that needs a migration the first time someone asks to share a view.\nRecommendation: SELECTIVE EXPANSION because this is an enhancement to an existing task list, not greenfield, and approach B already reserves room for the obvious expansions, so you should see them and decide rather than have me push.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nNet: how much of the 12-month ideal you want to look at now versus keep out of sight until the pilot reports back.",
|
||||
"options": [
|
||||
{
|
||||
"label": "Selective expansion (recommended)",
|
||||
"description": "\u2705 Baseline scope gets the full rigor pass: errors, edge cases, security, observability, rollout\n\u2705 Each expansion (shared views, default view, URL sync, view from filter) presented neutrally for cherry-pick\n\u274c More decision prompts than hold scope, roughly five or six extra questions"
|
||||
},
|
||||
{
|
||||
"label": "Scope expansion",
|
||||
"description": "\u2705 I describe the 10x version and platonic ideal first, then pitch each piece with enthusiasm\n\u2705 Best if you want the pilot to test the full vision, not the minimum\n\u274c Risks a larger first ship before the two-week pilot has any signal"
|
||||
},
|
||||
{
|
||||
"label": "Hold scope",
|
||||
"description": "\u2705 Fewest prompts; every minute goes to making the written scope bulletproof\n\u2705 Ships fastest for the pilot, expansions wait for pilot data\n\u274c You never see the add-on list, so a cheap win like URL sync could be missed"
|
||||
},
|
||||
{
|
||||
"label": "Scope reduction",
|
||||
"description": "\u2705 Strips to the absolute minimum that stops re-building filters, likely URL state plus last-used memory\n\u2705 Smallest security and migration surface\n\u274c Likely drops named views, which contradicts the goal and the approach you just chose"
|
||||
}
|
||||
]
|
||||
}
|
||||
]
|
||||
},
|
||||
"caller": {
|
||||
"type": "direct"
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
},
|
||||
{
|
||||
"type": "user",
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-H67mSL",
|
||||
"sessionId": "af7bf1c9-f26a-4b9b-96cc-02005d094575",
|
||||
"isSidechain": false,
|
||||
"timestamp": "2026-09-09T18:33:16.810Z",
|
||||
"message": {
|
||||
"role": "user",
|
||||
"content": [
|
||||
{
|
||||
"type": "tool_result",
|
||||
"content": "Your questions have been answered: \"D4 \u2014 Which review posture should I take for the rest of this review?\nProject/branch/task: gstack-plan-count-H67mSL on main, PLAN.md saved project views, approach B chosen.\nELI10: The approach is settled. Now I need to know how ambitious to be. Expansion means I dream up the cathedral version and pitch each piece to you one at a time. Selective expansion means I hold your scope as the baseline, make it bulletproof, and separately show you each add-on so you can cherry-pick with no sales pitch. Hold scope means no add-ons at all, just maximum rigor on what is written. Reduction means I cut the plan to the smallest thing that stops members re-building filters. Every scope change in every mode is still your explicit opt-in.\nStakes if we pick wrong: too ambitious and the pilot slips; too narrow and you ship a table that needs a migration the first time someone asks to share a view.\nRecommendation: SELECTIVE EXPANSION because this is an enhancement to an existing task list, not greenfield, and approach B already reserves room for the obvious expansions, so you should see them and decide rather than have me push.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nNet: how much of the 12-month ideal you want to look at now versus keep out of sight until the pilot reports back.\"=\"Hold scope\". You can now continue with these answers in mind.",
|
||||
"tool_use_id": "toolu_01LfRN9N3Fi92y3iQCREW1xT"
|
||||
}
|
||||
]
|
||||
},
|
||||
"toolUseResult": {
|
||||
"questions": [
|
||||
{
|
||||
"question": "D4 \u2014 Which review posture should I take for the rest of this review?\nProject/branch/task: gstack-plan-count-H67mSL on main, PLAN.md saved project views, approach B chosen.\nELI10: The approach is settled. Now I need to know how ambitious to be. Expansion means I dream up the cathedral version and pitch each piece to you one at a time. Selective expansion means I hold your scope as the baseline, make it bulletproof, and separately show you each add-on so you can cherry-pick with no sales pitch. Hold scope means no add-ons at all, just maximum rigor on what is written. Reduction means I cut the plan to the smallest thing that stops members re-building filters. Every scope change in every mode is still your explicit opt-in.\nStakes if we pick wrong: too ambitious and the pilot slips; too narrow and you ship a table that needs a migration the first time someone asks to share a view.\nRecommendation: SELECTIVE EXPANSION because this is an enhancement to an existing task list, not greenfield, and approach B already reserves room for the obvious expansions, so you should see them and decide rather than have me push.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nNet: how much of the 12-month ideal you want to look at now versus keep out of sight until the pilot reports back.",
|
||||
"header": "Review mode",
|
||||
"options": [
|
||||
{
|
||||
"label": "Selective expansion (recommended)",
|
||||
"description": "\u2705 Baseline scope gets the full rigor pass: errors, edge cases, security, observability, rollout\n\u2705 Each expansion (shared views, default view, URL sync, view from filter) presented neutrally for cherry-pick\n\u274c More decision prompts than hold scope, roughly five or six extra questions"
|
||||
},
|
||||
{
|
||||
"label": "Scope expansion",
|
||||
"description": "\u2705 I describe the 10x version and platonic ideal first, then pitch each piece with enthusiasm\n\u2705 Best if you want the pilot to test the full vision, not the minimum\n\u274c Risks a larger first ship before the two-week pilot has any signal"
|
||||
},
|
||||
{
|
||||
"label": "Hold scope",
|
||||
"description": "\u2705 Fewest prompts; every minute goes to making the written scope bulletproof\n\u2705 Ships fastest for the pilot, expansions wait for pilot data\n\u274c You never see the add-on list, so a cheap win like URL sync could be missed"
|
||||
},
|
||||
{
|
||||
"label": "Scope reduction",
|
||||
"description": "\u2705 Strips to the absolute minimum that stops re-building filters, likely URL state plus last-used memory\n\u2705 Smallest security and migration surface\n\u274c Likely drops named views, which contradicts the goal and the approach you just chose"
|
||||
}
|
||||
],
|
||||
"multiSelect": false
|
||||
}
|
||||
],
|
||||
"answers": {
|
||||
"D4 \u2014 Which review posture should I take for the rest of this review?\nProject/branch/task: gstack-plan-count-H67mSL on main, PLAN.md saved project views, approach B chosen.\nELI10: The approach is settled. Now I need to know how ambitious to be. Expansion means I dream up the cathedral version and pitch each piece to you one at a time. Selective expansion means I hold your scope as the baseline, make it bulletproof, and separately show you each add-on so you can cherry-pick with no sales pitch. Hold scope means no add-ons at all, just maximum rigor on what is written. Reduction means I cut the plan to the smallest thing that stops members re-building filters. Every scope change in every mode is still your explicit opt-in.\nStakes if we pick wrong: too ambitious and the pilot slips; too narrow and you ship a table that needs a migration the first time someone asks to share a view.\nRecommendation: SELECTIVE EXPANSION because this is an enhancement to an existing task list, not greenfield, and approach B already reserves room for the obvious expansions, so you should see them and decide rather than have me push.\nNote: options differ in kind, not coverage \u2014 no completeness score.\nNet: how much of the 12-month ideal you want to look at now versus keep out of sight until the pilot reports back.": "Hold scope"
|
||||
},
|
||||
"annotations": {}
|
||||
}
|
||||
},
|
||||
{
|
||||
"type": "assistant",
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-H67mSL",
|
||||
"sessionId": "af7bf1c9-f26a-4b9b-96cc-02005d094575",
|
||||
"isSidechain": false,
|
||||
"timestamp": "2026-09-09T18:35:04.651Z",
|
||||
"message": {
|
||||
"role": "assistant",
|
||||
"content": [
|
||||
{
|
||||
"type": "tool_use",
|
||||
"id": "toolu_01VmTaVZ1Jcdhodn8m3THdTr",
|
||||
"name": "AskUserQuestion",
|
||||
"input": {
|
||||
"questions": [
|
||||
{
|
||||
"header": "URL sync",
|
||||
"question": "D5 \u2014 Does the URL codec ship in this first release, or does it wait?\nProject/branch/task: gstack-plan-count-H67mSL on main, saved project views, HOLD SCOPE, approach B.\nELI10: Approach B defines one ViewState shape and said the URL would use it too. Writing the current filters into the URL means reload and browser back keep your filters, and a copied link carries them. Named saved views work without it. In HOLD SCOPE my job is to flag any work that can wait without blocking the goal, and this is the one piece that qualifies. It is not new scope either way; it is sequencing of an already approved approach.\nStakes if we pick wrong: ship it now and the first release has one more surface to test; wait and the pilot has no shareable link, so members send screenshots instead.\nRecommendation: A because the codec is the same serializer the API already needs, so the marginal cost is one encode/decode pair plus a router hook, roughly 10 minutes of CC time, and it removes the most common \"lost my filters on reload\" complaint the pilot would otherwise surface.\nCompleteness: A=10/10, B=7/10\nNet: about 10 minutes of CC time now versus a known pilot complaint and a follow-up PR later.",
|
||||
"options": [
|
||||
{
|
||||
"label": "A: Ship URL sync in v1 (recommended)",
|
||||
"description": "Completeness 10/10. Effort S (human ~half day / CC ~10 min).\n\u2705 Reload, back button, and copied links all preserve filters from day one of the pilot\n\u2705 Same serializer as the API, so one codec is tested once and reused twice\n\u2705 Page load has a clear precedence rule: URL state wins, then last-used view, then default\n\u274c One more surface to test: malformed or oversized URL state must degrade to the default list with a notice"
|
||||
},
|
||||
{
|
||||
"label": "B: Defer URL sync to follow-up",
|
||||
"description": "Completeness 7/10. Effort S later (human ~half day / CC ~10 min).\n\u2705 Smaller first release; picker and API are the only new surfaces to verify\n\u2705 Pilot feedback can confirm whether link sharing is actually wanted\n\u274c Members lose filters on reload during the pilot, which is the exact pain the plan targets"
|
||||
}
|
||||
]
|
||||
}
|
||||
]
|
||||
},
|
||||
"caller": {
|
||||
"type": "direct"
|
||||
}
|
||||
}
|
||||
]
|
||||
}
|
||||
},
|
||||
{
|
||||
"type": "user",
|
||||
"cwd": "/tmp/gstack-paid-shard-pa2Bc4/tmp/gstack-plan-count-H67mSL",
|
||||
"sessionId": "af7bf1c9-f26a-4b9b-96cc-02005d094575",
|
||||
"isSidechain": false,
|
||||
"timestamp": "2026-09-09T18:35:05.273Z",
|
||||
"message": {
|
||||
"role": "user",
|
||||
"content": [
|
||||
{
|
||||
"type": "tool_result",
|
||||
"content": "Your questions have been answered: \"D5 \u2014 Does the URL codec ship in this first release, or does it wait?\nProject/branch/task: gstack-plan-count-H67mSL on main, saved project views, HOLD SCOPE, approach B.\nELI10: Approach B defines one ViewState shape and said the URL would use it too. Writing the current filters into the URL means reload and browser back keep your filters, and a copied link carries them. Named saved views work without it. In HOLD SCOPE my job is to flag any work that can wait without blocking the goal, and this is the one piece that qualifies. It is not new scope either way; it is sequencing of an already approved approach.\nStakes if we pick wrong: ship it now and the first release has one more surface to test; wait and the pilot has no shareable link, so members send screenshots instead.\nRecommendation: A because the codec is the same serializer the API already needs, so the marginal cost is one encode/decode pair plus a router hook, roughly 10 minutes of CC time, and it removes the most common \"lost my filters on reload\" complaint the pilot would otherwise surface.\nCompleteness: A=10/10, B=7/10\nNet: about 10 minutes of CC time now versus a known pilot complaint and a follow-up PR later.\"=\"A: Ship URL sync in v1 (recommended)\". You can now continue with these answers in mind.",
|
||||
"tool_use_id": "toolu_01VmTaVZ1Jcdhodn8m3THdTr"
|
||||
}
|
||||
]
|
||||
},
|
||||
"toolUseResult": {
|
||||
"questions": [
|
||||
{
|
||||
"question": "D5 \u2014 Does the URL codec ship in this first release, or does it wait?\nProject/branch/task: gstack-plan-count-H67mSL on main, saved project views, HOLD SCOPE, approach B.\nELI10: Approach B defines one ViewState shape and said the URL would use it too. Writing the current filters into the URL means reload and browser back keep your filters, and a copied link carries them. Named saved views work without it. In HOLD SCOPE my job is to flag any work that can wait without blocking the goal, and this is the one piece that qualifies. It is not new scope either way; it is sequencing of an already approved approach.\nStakes if we pick wrong: ship it now and the first release has one more surface to test; wait and the pilot has no shareable link, so members send screenshots instead.\nRecommendation: A because the codec is the same serializer the API already needs, so the marginal cost is one encode/decode pair plus a router hook, roughly 10 minutes of CC time, and it removes the most common \"lost my filters on reload\" complaint the pilot would otherwise surface.\nCompleteness: A=10/10, B=7/10\nNet: about 10 minutes of CC time now versus a known pilot complaint and a follow-up PR later.",
|
||||
"header": "URL sync",
|
||||
"options": [
|
||||
{
|
||||
"label": "A: Ship URL sync in v1 (recommended)",
|
||||
"description": "Completeness 10/10. Effort S (human ~half day / CC ~10 min).\n\u2705 Reload, back button, and copied links all preserve filters from day one of the pilot\n\u2705 Same serializer as the API, so one codec is tested once and reused twice\n\u2705 Page load has a clear precedence rule: URL state wins, then last-used view, then default\n\u274c One more surface to test: malformed or oversized URL state must degrade to the default list with a notice"
|
||||
},
|
||||
{
|
||||
"label": "B: Defer URL sync to follow-up",
|
||||
"description": "Completeness 7/10. Effort S later (human ~half day / CC ~10 min).\n\u2705 Smaller first release; picker and API are the only new surfaces to verify\n\u2705 Pilot feedback can confirm whether link sharing is actually wanted\n\u274c Members lose filters on reload during the pilot, which is the exact pain the plan targets"
|
||||
}
|
||||
],
|
||||
"multiSelect": false
|
||||
}
|
||||
],
|
||||
"answers": {
|
||||
"D5 \u2014 Does the URL codec ship in this first release, or does it wait?\nProject/branch/task: gstack-plan-count-H67mSL on main, saved project views, HOLD SCOPE, approach B.\nELI10: Approach B defines one ViewState shape and said the URL would use it too. Writing the current filters into the URL means reload and browser back keep your filters, and a copied link carries them. Named saved views work without it. In HOLD SCOPE my job is to flag any work that can wait without blocking the goal, and this is the one piece that qualifies. It is not new scope either way; it is sequencing of an already approved approach.\nStakes if we pick wrong: ship it now and the first release has one more surface to test; wait and the pilot has no shareable link, so members send screenshots instead.\nRecommendation: A because the codec is the same serializer the API already needs, so the marginal cost is one encode/decode pair plus a router hook, roughly 10 minutes of CC time, and it removes the most common \"lost my filters on reload\" complaint the pilot would otherwise surface.\nCompleteness: A=10/10, B=7/10\nNet: about 10 minutes of CC time now versus a known pilot complaint and a follow-up PR later.": "A: Ship URL sync in v1 (recommended)"
|
||||
},
|
||||
"annotations": {}
|
||||
}
|
||||
}
|
||||
],
|
||||
"provenance": "Exact prior Approach B approval plus mode and completed URL sequencing decision. Retained only public AUQ records; no scope-approval broker inferred."
|
||||
}
|
||||
]
|
||||
}
|
||||
+98
@@ -0,0 +1,98 @@
|
||||
{
|
||||
"provenance": {
|
||||
"run": "ship-source-ad-full-paid-20260909-v1",
|
||||
"sourceHead": "d4f42b69a1dd0eb48d89500e03fb1c4407229284",
|
||||
"captureProofSHA256": "62abe34ea7f79f91af352bee36a18ce5fe7fbef1dee84f5427214a626ab5bcf2",
|
||||
"screens": "Exact retained current viewports; original raw hook state is content-free; transcript projection retains only status/session identity. These three calls were pending at operator cancellation and never award completion or review credit."
|
||||
},
|
||||
"rows": [
|
||||
{
|
||||
"pid": 1332470,
|
||||
"cwd": "/tmp/gstack-paid-shard-dLDXE6/tmp/gstack-plan-count-9QMoxP",
|
||||
"config": "/tmp/gstack-paid-shard-dLDXE6/tmp/gstack-hermetic-1332375-n7QqpR/with-skills/.claude",
|
||||
"startUnix": 1788977964.97,
|
||||
"screen": " 136 + | | (name/email, digest on, tips off)| \n + | | \n 137 +Save | PENDING (Gap 2): currently | n/a | Validation: beside field + l\n +inked summary, | InlineStatus “Saved at HH:mm” | Never exposed: save is atomic \n 138 + | no indicator for 2-5s | | focus first invalid field. N\n +etwork: edits | (24h local), Export re-enabled | \n 139 + | | | kept, “Unsaved changes” + er\n +ror + Retry | | \n 140 + | | | (aria-label “Retry save”). C\n +olor PENDING | | \n 141 + | | | (Gap 3). \n + | | \n 142 +Dirty state | n/a | Before first save: blank status | n/a \n + | Revert/Reset restores “Saved | n/a \n 143 + | | | \n + | at HH:mm” or blank pre-save | \n 144 +Reset | n/a | Disabled/no-op when clean | n/a \n + | Dialog confirm → saved values, | n/a \n 145 + | | | \n + | focus back to Reset trigger | \n 146 +Cancel | n/a | Clean: navigate back directly | n/a \n + | Dirty: dialog → “Discard and | n/a \n 147 + | | | \n + | leave” → destination h1 focus | \n 148 +Export | Both Save and Export | n/a (defaults always exist) | Same inline error/retry area\n +, fields kept, | JSON download of last saved | n/a \n 149 + | disabled while pending | | Retry aria-label “Retry expo\n +rt” | preferences; clears only the | \n 150 + | | | \n + | Export error | \n 151 +``` \n╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌\n Do you want to make this edit to gstack-test-plan-design.md?\n ❯ 1. Yes\n 2. Yes, and switch to accept edits (auto-approve file edits and common file commands) for this session; Yes, and\n always allow access to /tmp/gstack-paid-shard-dLDXE6/tmp/gstack-e2e-plan-design-oxarVN for this session\n 3. Nohift+tab)\n\n Esc to cancel · Tab to amend\n",
|
||||
"screenSHA256": "cfe0ec3902cb0f989d3b264d43bc1ea5b568fb5595e335ce95869ec497465548",
|
||||
"state": {
|
||||
"cwd": "/tmp/gstack-paid-shard-dLDXE6/tmp/gstack-plan-count-9QMoxP",
|
||||
"expected": "/tmp/gstack-paid-shard-dLDXE6/tmp/gstack-e2e-plan-design-oxarVN/gstack-test-plan-design.md",
|
||||
"sessionId": "4d9db343-ed09-40d4-9813-6a865ebf5dfc",
|
||||
"transcriptPath": "/tmp/gstack-paid-shard-dLDXE6/tmp/gstack-hermetic-1332375-n7QqpR/with-skills/.claude/projects/-tmp-gstack-paid-shard-dLDXE6-tmp-gstack-plan-count-9QMoxP/4d9db343-ed09-40d4-9813-6a865ebf5dfc.jsonl",
|
||||
"seenIds": [
|
||||
"4d9db343-ed09-40d4-9813-6a865ebf5dfc:toolu_01XJjXS1coVMTm9HF7tGU8hs",
|
||||
"4d9db343-ed09-40d4-9813-6a865ebf5dfc:toolu_01Qb8PpdQK7z7Xu3ESFv757b",
|
||||
"4d9db343-ed09-40d4-9813-6a865ebf5dfc:toolu_0195ubwPQhDuzoa3WkqLbnsG"
|
||||
],
|
||||
"pendingId": "4d9db343-ed09-40d4-9813-6a865ebf5dfc:toolu_0195ubwPQhDuzoa3WkqLbnsG",
|
||||
"completedId": "4d9db343-ed09-40d4-9813-6a865ebf5dfc:toolu_01Qb8PpdQK7z7Xu3ESFv757b",
|
||||
"completedIds": [
|
||||
"4d9db343-ed09-40d4-9813-6a865ebf5dfc:toolu_01XJjXS1coVMTm9HF7tGU8hs",
|
||||
"4d9db343-ed09-40d4-9813-6a865ebf5dfc:toolu_01Qb8PpdQK7z7Xu3ESFv757b"
|
||||
],
|
||||
"timestamp": "2026-09-09T18:23:13.771Z"
|
||||
},
|
||||
"transcriptStatus": "ready",
|
||||
"transcriptSessions": [
|
||||
"4d9db343-ed09-40d4-9813-6a865ebf5dfc"
|
||||
]
|
||||
},
|
||||
{
|
||||
"pid": 1332996,
|
||||
"cwd": "/tmp/gstack-paid-shard-iA05Vy/tmp/gstack-plan-count-21Yj95",
|
||||
"config": "/tmp/gstack-paid-shard-iA05Vy/tmp/gstack-hermetic-1332950-AN120U/with-skills/.claude",
|
||||
"startUnix": 1788977969.01,
|
||||
"screen": " +e retries OK \n 100 +handler#send_notification provider timeout / 5xx / SMTP mail client's named Y (D5) nothing; ale\n +rt + runbook FIXED \n 101 + delivery exceptions \n 102 +handler#send_notification fired before commit (ordering) n/a dup email on\n + retry FIXED (D5) \n 103 +handler#send_notification mail outage inside the txn mail client error n/a payment neve\n +r recorded FIXED (D5) \n 104 +``` \n 105 + \n 106 +Finding 2.1 — Email leg had undefined ordering and no rescue. **Decision D5 = A:** \n 107 +lookup → update inside the transaction → commit → dedup marks complete → send \n 108 +email. Rescue ONLY the mail client's named delivery exceptions (timeout, \n 109 +delivery failure; never StandardError). On rescue: error-level log with event \n 110 +ID, user ID, payment intent ID, exception class; the mail client's existing \n 111 +failure metric and on-call alert fire; return 200. Any other exception \n 112 +propagates unchanged. Tests: post-commit ordering, rescued mail failure returns \n 113 +200 and logs, non-mail exception propagates. \n 114 + \n 115 +## Section 3: Security & Threat Model \n 116 + \n 117 +| Threat | Likelihood | Impact | Mitigated? | \n 118 +|---|---|---|---| \n 119 +| SQL injection via `request.params.userId` in a raw SQL fragment | Med | High | No (see 3.1) | \n 120 +| Forged webhook | Low | High | Yes: existing signature middleware | \n 121 +| Replayed event | Low | Low | Yes: event-ID dedup; update is idempotent assignment | \n 122 +| Attacker-influenced metadata marks the wrong user paid | Low | High | Outside this handler; belongs to the check\n +out flow that sets metadata (NOT in scope) | \n 123 +| Card data / PII | n/a | n/a | No card data; IDs already traced by existing clients | \n 124 + \n 125 _(review in progress — sections appended as decisions land)_\n╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌\n Do you want to make this edit to gstack-test-plan-ceo.md?\n ❯ 1. Yes\n 2. Yes, and switch to accept edits (auto-approve file edits and common file commands) for this session; Yes, and\n always allow access to /tmp/gstack-paid-shard-iA05Vy/tmp/gstack-e2e-plan-ceo-I6YhZN for this session (shift+tab)\n 3. No\n\n Esc to cancel · Tab to amend\n",
|
||||
"screenSHA256": "d87280407169d000329f7567a982b14b0529b94db5e2758c9587e805f9af9bac",
|
||||
"state": {
|
||||
"cwd": "/tmp/gstack-paid-shard-iA05Vy/tmp/gstack-plan-count-21Yj95",
|
||||
"expected": "/tmp/gstack-paid-shard-iA05Vy/tmp/gstack-e2e-plan-ceo-I6YhZN/gstack-test-plan-ceo.md",
|
||||
"sessionId": "3e71b5ab-fec6-4cce-ba20-9d60b3b9970d",
|
||||
"transcriptPath": "/tmp/gstack-paid-shard-iA05Vy/tmp/gstack-hermetic-1332950-AN120U/with-skills/.claude/projects/-tmp-gstack-paid-shard-iA05Vy-tmp-gstack-plan-count-21Yj95/3e71b5ab-fec6-4cce-ba20-9d60b3b9970d.jsonl",
|
||||
"seenIds": [
|
||||
"3e71b5ab-fec6-4cce-ba20-9d60b3b9970d:toolu_01J4snDFPcLUS4dtBx3wPdNv",
|
||||
"3e71b5ab-fec6-4cce-ba20-9d60b3b9970d:toolu_01ES9kGmZCiY7ZXMF1SMig2g",
|
||||
"3e71b5ab-fec6-4cce-ba20-9d60b3b9970d:toolu_011t5RMqaJ72mFr761edaw9e"
|
||||
],
|
||||
"pendingId": "3e71b5ab-fec6-4cce-ba20-9d60b3b9970d:toolu_011t5RMqaJ72mFr761edaw9e",
|
||||
"completedId": "3e71b5ab-fec6-4cce-ba20-9d60b3b9970d:toolu_01ES9kGmZCiY7ZXMF1SMig2g",
|
||||
"completedIds": [
|
||||
"3e71b5ab-fec6-4cce-ba20-9d60b3b9970d:toolu_01J4snDFPcLUS4dtBx3wPdNv",
|
||||
"3e71b5ab-fec6-4cce-ba20-9d60b3b9970d:toolu_01ES9kGmZCiY7ZXMF1SMig2g"
|
||||
],
|
||||
"timestamp": "2026-09-09T18:25:17.883Z"
|
||||
},
|
||||
"transcriptStatus": "ready",
|
||||
"transcriptSessions": [
|
||||
"3e71b5ab-fec6-4cce-ba20-9d60b3b9970d"
|
||||
]
|
||||
},
|
||||
{
|
||||
"pid": 1333364,
|
||||
"cwd": "/tmp/gstack-paid-shard-ZHj9Ga/tmp/gstack-plan-count-F6E4PY",
|
||||
"config": "/tmp/gstack-paid-shard-ZHj9Ga/tmp/gstack-hermetic-1333312-CatzTs/with-skills/.claude",
|
||||
"startUnix": 1788977973.31,
|
||||
"screen": " +t path still reports `EVALKIT_CI_TIMEOUT`. \n 304 +5. Run a v1 script using `Client.evaluate()` and `run_batch(evaluator, dataset)`; expect two DeprecationWarnings a\n +nd correct results. \n 305 +6. Re-run the onboarding timing (T6); record the result in docs/benchmarks.md. \n 306 + \n 307 +## Deferred housekeeping (outside plan mode) \n 308 +- Append the gstack \"## Skill routing\" section to CLAUDE.md and commit it (user chose A on the routing prompt; def\n +erred because plan mode forbids the edit). \n 309 +- Cross-project learnings preference was not asked: the learnings store is empty in this environment, so the toggl\n +e has no effect yet. It will be offered on the next run. \n 310 +- Recommended next review: `/plan-eng-review` to validate the runtime changes in T1-T3 and T5, then `/devex-review\n +` on the shipped wheel to check the sub-2-minute claim. \n 311 + \n 312 +## GSTACK REVIEW REPORT \n 313 + \n 314 +| Review | Trigger | Why | Runs | Status | Findings | \n 315 +|--------|---------|-----|------|--------|----------| \n 316 +| CEO Review | `/plan-ceo-review` | Scope & strategy | 0 | — | — | \n 317 +| Outside Review | codex via `/plan-devex-review` | Independent 2nd opinion | 1 | disabled (skipped) | none, codex\n +_reviews disabled | \n 318 +| Eng Review | `/plan-eng-review` | Architecture & tests (required) | 0 | — | — | \n 319 +| Design Review | `/plan-design-review` | UI/UX gaps | 0 | — | — | \n 320 +| DX Review | `/plan-devex-review` | Developer experience gaps | 1 | clean | score: 5/10 → 9/10, TTHW: 6 min → <2 \n +min (est.), 5 findings, 5 decisions, mode POLISH | \n 321 + \n 322 +**OUTSIDE COVERAGE:** provider codex, phase plan-review, outside_status disabled (user config `codex_reviews=disab\n +led`), no findings; native fallback intentionally not dispatched. \n 323 + \n 324 +**VERDICT:** DX CLEARED — eng review required (no `/plan-eng-review` or `/review` entry in the last 7 days). \n 325 + \n 326 +NO UNRESOLVED DECISIONS \n╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌\n Do you want to make this edit to gstack-test-plan-devex.md?\n ❯ 1. Yes\n 2. Yes, and switch to accept edits (auto-approve file edits and common file commands) for this session; Yes, and\n always allow access to /tmp/gstack-paid-shard-ZHj9Ga/tmp/gstack-e2e-plan-devex-Pn1BeW for this session (shift+tab)\n 3. No\n\n Esc to cancel · Tab to amend\n",
|
||||
"screenSHA256": "31548208a094334e0072323081570cbdcd3529833bc5e803cadc56466c4227dd",
|
||||
"state": {
|
||||
"cwd": "/tmp/gstack-paid-shard-ZHj9Ga/tmp/gstack-plan-count-F6E4PY",
|
||||
"expected": "/tmp/gstack-paid-shard-ZHj9Ga/tmp/gstack-e2e-plan-devex-Pn1BeW/gstack-test-plan-devex.md",
|
||||
"sessionId": "40207c5f-0ce9-4ec9-b09d-ec0ba8f93089",
|
||||
"transcriptPath": "/tmp/gstack-paid-shard-ZHj9Ga/tmp/gstack-hermetic-1333312-CatzTs/with-skills/.claude/projects/-tmp-gstack-paid-shard-ZHj9Ga-tmp-gstack-plan-count-F6E4PY/40207c5f-0ce9-4ec9-b09d-ec0ba8f93089.jsonl",
|
||||
"seenIds": [
|
||||
"40207c5f-0ce9-4ec9-b09d-ec0ba8f93089:toolu_01QLo6kZHJWRkGmtn92aBW81",
|
||||
"40207c5f-0ce9-4ec9-b09d-ec0ba8f93089:toolu_018vYkQ6sYgziye4GfXFyWDF"
|
||||
],
|
||||
"pendingId": "40207c5f-0ce9-4ec9-b09d-ec0ba8f93089:toolu_018vYkQ6sYgziye4GfXFyWDF",
|
||||
"completedId": "40207c5f-0ce9-4ec9-b09d-ec0ba8f93089:toolu_01QLo6kZHJWRkGmtn92aBW81",
|
||||
"completedIds": [
|
||||
"40207c5f-0ce9-4ec9-b09d-ec0ba8f93089:toolu_01QLo6kZHJWRkGmtn92aBW81"
|
||||
],
|
||||
"timestamp": "2026-09-09T18:26:53.932Z"
|
||||
},
|
||||
"transcriptStatus": "ready",
|
||||
"transcriptSessions": [
|
||||
"40207c5f-0ce9-4ec9-b09d-ec0ba8f93089"
|
||||
]
|
||||
}
|
||||
]
|
||||
}
|
||||
@@ -6,9 +6,9 @@ export interface AutoplanPhaseHit {
|
||||
}
|
||||
|
||||
function phaseDeclaration(text: string): RegExpExecArray | null {
|
||||
const plain = text.replace(/^\*\*(Phase[ \t]+[\d.]+[ \t]+complete\.?)\*\*/i, '$1');
|
||||
const plain = text.replace(/^\*\*(Phase[ \t]+[\d.]+[ \t]+(?:complete|is[ \t]+done)[.:]?)\*\*/i, '$1');
|
||||
if (/\bEmit\s+phase-transition\s+summary\s*:/i.test(plain)) return null;
|
||||
return /^Phase[ \t]+(1|2(?:\.5)?|3)[ \t]+complete(?:\.(?:[ \t]+.*)?|)$/i.exec(plain);
|
||||
return /^Phase[ \t]+(1|2(?:\.5)?|3)[ \t]+(?:complete(?:\.(?:[ \t]+.*)?|)|is[ \t]+done(?:[.:](?:[ \t]+.*)?|))$/i.exec(plain);
|
||||
}
|
||||
|
||||
/**
|
||||
|
||||
@@ -134,6 +134,18 @@ export function hasPostAnswerCeoPosture(visible: string, posture: RegExp): boole
|
||||
return matches();
|
||||
}
|
||||
|
||||
/** Apply the same echo/quotation exclusions to native prose and decision rationale. */
|
||||
function hasNativePostureProse(text: string, posture: RegExp): boolean {
|
||||
const prose = text.replace(/```[\s\S]*?```/g, '').split('\n').filter(line => {
|
||||
if (/^\s*>/.test(line)) return false;
|
||||
const plain = line.replace(/[*_`]/g, '').trim().replace(/^#+\s*/, '');
|
||||
// A repeated menu or bare confirmation is still only an answer echo.
|
||||
if (/^(?:[-+]|\d+[.)]|[A-D][.)])\s*(?:HOLD SCOPE|SCOPE EXPANSION|SELECTIVE EXPANSION|SCOPE REDUCTION)\b/i.test(plain)) return false;
|
||||
return !/^(?:(?:You\s+)?selected(?:\s+(?:option|mode))?\s*[:：]?\s*)?(?:HOLD SCOPE|SCOPE EXPANSION|SELECTIVE EXPANSION|SCOPE REDUCTION)(?:\s+mode)?(?:\s+confirmed)?(?:\s*\(recommended\))?[.!]?$/i.test(plain);
|
||||
}).join('\n');
|
||||
return hasPostAnswerCeoPosture(`● ${prose}`, posture);
|
||||
}
|
||||
|
||||
/** A native successful answer, not the key we intended to send to the menu. */
|
||||
export function nativeCeoModeAnswer(
|
||||
transcript: PlanCountTranscript,
|
||||
@@ -176,7 +188,7 @@ function completedQuestionTimes(call: NativePlanQuestionCall, events: ReadonlyAr
|
||||
}
|
||||
|
||||
/** New shorthand forms must be one complete decision, not a mode mention or extra question. */
|
||||
function singleScopeBrief(text: string, descriptions: readonly string[]): boolean {
|
||||
function singleScopeBrief(text: string, descriptions: readonly string[], comparison = true): boolean {
|
||||
if ([text, ...descriptions].some(value => /(?:^|[.!?]\s+|\n)\s*(?:Also|Separately|Additionally)\b|\b(?:Please|We must|You must|The plan must)\b/i.test(value))) return false;
|
||||
// Query parameter names such as ?view= are not another decision prompt.
|
||||
const questions = text.replace(/\?[A-Za-z_][\w-]*=/g, '=').match(/\?/g);
|
||||
@@ -192,7 +204,9 @@ function singleScopeBrief(text: string, descriptions: readonly string[]): boolea
|
||||
});
|
||||
// Net closes this decision brief. A following instruction is not part of its
|
||||
// comparison; this is not a general classifier of instructions inside prose.
|
||||
return complete && /^[^.!?;\n]+ (?:vs|versus) [^.!?;\n]+\.$/.test(text.slice(previous + 'Net:'.length).trim());
|
||||
const net = text.slice(previous + 'Net:'.length).trim();
|
||||
return complete && (comparison ? /^[^.!?;\n]+ (?:vs|versus) [^.!?;\n]+\.$/.test(net)
|
||||
: /^[^.!?;\n]+\.$/.test(net.replace(/\bvs\./gi, 'vs')));
|
||||
}
|
||||
|
||||
/** HOLD can apply its boundary in a completed defer decision, before standalone prose. */
|
||||
@@ -205,6 +219,17 @@ function hasAnsweredHoldPosture(transcript: PlanCountTranscript, selected: Nativ
|
||||
const times = completedQuestionTimes(call, events);
|
||||
if (!times || times.requestedAt <= modeTimes.answeredAt) return false;
|
||||
const q = call.questions[0]!;
|
||||
// A substantive review decision can apply HOLD in its rationale before
|
||||
// standalone prose is published. Metadata and answer echoes do not count.
|
||||
// This recognizes posture language; it does not validate every scope choice.
|
||||
const context = /Project\/branch\/task:([\s\S]*?)(?=ELI10:)/i.exec(q.question)?.[1] ?? '';
|
||||
const rationale = /ELI10:([\s\S]*?)(?=Stakes if (?:we pick )?wrong:)/i.exec(q.question)?.[1]?.trim() ?? '';
|
||||
const offered = q.options.map(o => o.label.trim());
|
||||
if (!q.multiSelect && q.options.length >= 2 && q.options.length <= 4 && new Set(offered).size === offered.length &&
|
||||
offered.includes(call.answers?.[q.question] ?? '') && /\bHOLD SCOPE\b/i.test(context) &&
|
||||
!/\b(?:SCOPE EXPANSION|SELECTIVE EXPANSION|SCOPE REDUCTION)\b/i.test(context) &&
|
||||
singleScopeBrief(q.question, q.options.map(o => o.description ?? ''), false) &&
|
||||
hasNativePostureProse(rationale, posture)) return true;
|
||||
if (q.multiSelect || q.options.length !== 3 || !singleScopeBrief(q.question, q.options.map(o => o.description ?? ''))) return false;
|
||||
const title = /^D\d+\s*[—–-]\s*Under HOLD SCOPE, keep or defer the (\w+)\b([\s\S]+?)\?\s+Project\/branch\/task:/i.exec(q.question);
|
||||
if (!title || !/\bnot in the plan text\b/i.test(title[2]!) ||
|
||||
@@ -234,16 +259,20 @@ function hasAnsweredExpansionPosture(
|
||||
const times = completedQuestionTimes(call, events);
|
||||
if (!times || times.requestedAt <= modeTimes.answeredAt) return false;
|
||||
const question = call.questions[0]!;
|
||||
// This is evidence that the selected mode produced a concrete scope
|
||||
// decision, not authority to answer it. Numbering and heading names vary.
|
||||
const title = question.question.split('\n')[0]!
|
||||
.replace(/\s*<gstack-qid:plan-ceo-review-expansion-[a-z0-9-]+>\s*$/i, '');
|
||||
const numbered = /^D\d+\s*[—–-]\s*Expansion\s+\d+\s+of\s+\d+:\s+\S.+\?\s*$/i.test(title);
|
||||
const short = /^D\d+\s*[—–-]\s*E(\d+):\s+Add\s+[^?\n]+\?\s*$/i.exec(title);
|
||||
const explicitShort = short && new RegExp(`^E${short[1]}\\s+\\S`, 'i').test(question.header) &&
|
||||
/\nProject\/branch\/task:[^\n]+\bSCOPE EXPANSION mode\.\s*\nELI10:/i.test(question.question) &&
|
||||
singleScopeBrief(question.question, question.options.map(o => o.description ?? ''));
|
||||
if (question.multiSelect || question.options.length !== 3 || (!numbered && !explicitShort)) return false;
|
||||
.replace(/\s*<gstack-qid:[a-z0-9-]+>\s*$/i, '').replace(/^D\d+\s*[—–-]\s*/i, '');
|
||||
const context = /\nProject\/branch\/task:([^\n]+)/i.exec(question.question)?.[1] ?? '';
|
||||
if (question.multiSelect || question.options.length !== 3 ||
|
||||
!/^[\p{L}\p{N}][^?\n]+\?$/u.test(title) ||
|
||||
/\b(?:review\s+(?:mode|posture)|(?:selected|confirmed)\s+(?:mode|option))\b/i.test(title) ||
|
||||
/\b(?:HOLD SCOPE|SELECTIVE EXPANSION|SCOPE REDUCTION)\b/i.test(context) ||
|
||||
!/\b(?:SCOPE\s+EXPANSION|EXPANSION\s+(?:mode|opt[ -]in))\b/i.test(context) ||
|
||||
!singleScopeBrief(question.question, question.options.map(o => o.description ?? ''), false)) return false;
|
||||
const labels = question.options.map(option => option.label.trim()
|
||||
.replace(/^[A-C][):.]\s*/i, '').replace(/\s*\(recommended\)\s*$/i, '').toLowerCase());
|
||||
.replace(/^[A-C][):.]\s*/i, '').replace(/\s*\(recommended\)\s*$/i, '').toLowerCase()
|
||||
.replace(/^add to (?:(?:this|the) plan['’]s )?scope$/, 'add to scope'));
|
||||
if (new Set(labels).size !== 3 || !['add to scope', 'defer to todos.md', 'skip'].every(label => labels.includes(label)) ||
|
||||
!question.options.some(option => option.label === call.answers?.[question.question])) return false;
|
||||
// Never search quoted instructions, tool output or a menu for posture.
|
||||
@@ -266,14 +295,7 @@ export function hasNativePostAnswerCeoPosture(
|
||||
const answeredAt = Date.parse(selected.answeredAt!);
|
||||
return transcript.assistantMessages.some(message => {
|
||||
if (message.sessionId !== selected.sessionId || Date.parse(message.timestamp) <= answeredAt) return false;
|
||||
const prose = message.text.replace(/```[\s\S]*?```/g, '').split('\n').filter(line => {
|
||||
if (/^\s*>/.test(line)) return false;
|
||||
const plain = line.replace(/[*_`]/g, '').trim().replace(/^#+\s*/, '');
|
||||
// A repeated menu or bare confirmation is still only an answer echo.
|
||||
if (/^(?:[-+]|\d+[.)]|[A-D][.)])\s*(?:HOLD SCOPE|SCOPE EXPANSION|SELECTIVE EXPANSION|SCOPE REDUCTION)\b/i.test(plain)) return false;
|
||||
return !/^(?:(?:You\s+)?selected(?:\s+(?:option|mode))?\s*[:：]?\s*)?(?:HOLD SCOPE|SCOPE EXPANSION|SELECTIVE EXPANSION|SCOPE REDUCTION)(?:\s+mode)?(?:\s+confirmed)?(?:\s*\(recommended\))?[.!]?$/i.test(plain);
|
||||
}).join('\n');
|
||||
return hasPostAnswerCeoPosture(`● ${prose}`, posture);
|
||||
return hasNativePostureProse(message.text, posture);
|
||||
}) || (targetMode === 'SCOPE EXPANSION' && hasAnsweredExpansionPosture(transcript, selected, posture, publicTools)) ||
|
||||
(targetMode === 'HOLD SCOPE' && hasAnsweredHoldPosture(transcript, selected, posture, publicTools));
|
||||
}
|
||||
|
||||
@@ -1561,21 +1561,29 @@ export function planCountPrerequisitePick(fp: AskUserQuestionFingerprint, active
|
||||
const skip = fp.options.filter(({ label }) =>
|
||||
/^Skip\s*[—–-]\s*(?:proceed\s*with\s*)?standard\s*review(?:\s*\(recommended\))?$/i.test(label));
|
||||
if (run.length === 1 && skip.length === 1) return skip[0].index;
|
||||
// Short labels such as "Skip — proceed" need their full native meaning.
|
||||
// A visible generic Skip alone cannot authorize leaving a review question.
|
||||
// Short labels need the complete meaning of the active native tab. Other
|
||||
// questions in a packet cannot lend their prerequisite context or choices.
|
||||
const call = activeCapture.nativeCall;
|
||||
if (call && call.answered === false && call.failed === false && activeCapture.preReview && call.questions.length === 1 &&
|
||||
activeCapture.signature === `${call.sessionId}:${call.toolUseId}` &&
|
||||
fp.signature === activeCapture.signature &&
|
||||
(activeCapture.nativeQuestionIndex === undefined || activeCapture.nativeQuestionIndex === 0)) {
|
||||
const q = call.questions[0]!;
|
||||
const activeIndex = activeCapture.nativeQuestionIndex ?? (call?.questions.length === 1 ? 0 : -1);
|
||||
if (call && call.answered === false && call.failed === false && activeCapture.preReview &&
|
||||
Number.isInteger(activeIndex) && activeIndex >= 0 && activeIndex < call.questions.length) {
|
||||
const q = call.questions[activeIndex]!;
|
||||
const base = `${call.sessionId}:${call.toolUseId}`;
|
||||
const activeSignature = call.questions.length === 1 ? base : `${base}:question:${activeIndex}`;
|
||||
const routing = nativePlanCallFingerprint(call, fp.observedAtMs, fp.preReview);
|
||||
const optionsMatch = (capture: AskUserQuestionFingerprint) => capture.options.length === q.options.length &&
|
||||
capture.options.every((option, i) => option.index === i + 1 && option.label === q.options[i]!.label);
|
||||
if (!q.multiSelect && q.options.length === 2 && optionsMatch(fp) && optionsMatch(activeCapture)) {
|
||||
const routingMatches = fp.nativeCall === call && (fp.signature === activeSignature
|
||||
? fp.promptSnippet === `${q.header} ${q.question}` && optionsMatch(fp)
|
||||
: fp.signature === base && fp.promptSnippet === routing.promptSnippet &&
|
||||
JSON.stringify(fp.options) === JSON.stringify(routing.options));
|
||||
if (!q.multiSelect && q.options.length === 2 && activeCapture.signature === activeSignature &&
|
||||
activeCapture.promptSnippet === `${q.header} ${q.question}` && optionsMatch(activeCapture) && routingMatches &&
|
||||
/\/office-hours/i.test(q.question) && /(?:no\s*design\s*doc|produce\s*a\s*design\s*doc)/i.test(q.question)) {
|
||||
const nativeRun = q.options.findIndex(option => /^Run\s*\/office-hours\s*(?:now|first)(?:\s*\(recommended\))?$/i.test(option.label));
|
||||
const nativeSkip = q.options.findIndex(option =>
|
||||
/^Skip\s*[—–-]\s*proceed(?:\s*\(recommended\))?$/i.test(option.label) &&
|
||||
/^(?:(?:The\s+)?plan\s+(?:scope\s+)?is\s+(?:already\s+)?(?:precise|clear|well-defined|explicit)\.\s*)?Proceed\s+with\s+standard(?:\s+DX(?:\s+(?:POLISH|EXPANSION|TRIAGE))?)?\s+review\.?$/i.test((option.description ?? '').trim()));
|
||||
/^Skip(?:\s*[—–-]\s*proceed)?(?:\s*\(recommended\))?$/i.test(option.label) &&
|
||||
/^(?:(?:The\s+)?plan\s+(?:scope\s+)?is\s+(?:already\s+)?(?:precise|clear|well-defined|explicit)\.\s*)?Proceed\s+(?:with\s+standard(?:\s+DX(?:\s+(?:POLISH|EXPANSION|TRIAGE))?)?\s+review|straight\s+to\s+Step\s*0\s+premise\s+challenge\s+and\s+approach\s+alternatives)\.?$/i.test((option.description ?? '').trim()));
|
||||
if (nativeRun >= 0 && nativeSkip >= 0 && nativeRun !== nativeSkip) return nativeSkip + 1;
|
||||
}
|
||||
}
|
||||
@@ -2467,9 +2475,10 @@ export async function launchClaudePty(
|
||||
if (opts.env?.HOME === undefined) {
|
||||
const runtime = withHermeticSkillRuntime(childEnv);
|
||||
childEnv = runtime.env;
|
||||
// Lazy sections are installed runtime inputs outside the fixture cwd.
|
||||
// Grant this known directory, preserving all other permission decisions.
|
||||
args.push('--add-dir', runtime.root);
|
||||
// Installed sections and generated ~/.gstack snapshots are owned inputs
|
||||
// outside the fixture cwd. Grant only these two runtime directories;
|
||||
// explicit GSTACK_HOME paths and operator permission settings stay separate.
|
||||
args.push('--add-dir', runtime.root, '--add-dir', runtime.stateRoot);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
@@ -16,13 +16,17 @@ import * as path from 'node:path';
|
||||
import * as os from 'node:os';
|
||||
import { getHermeticDirs, hermeticSkillsConfigDir } from './hermetic-env';
|
||||
|
||||
let cachedRuntime: { home: string; root: string } | undefined;
|
||||
let cachedRuntime: { home: string; root: string; stateRoot: string } | undefined;
|
||||
|
||||
export function hermeticSkillRuntime(): { home: string; root: string } {
|
||||
export function hermeticSkillRuntime(): { home: string; root: string; stateRoot: string } {
|
||||
if (cachedRuntime) return cachedRuntime;
|
||||
const home = fs.mkdtempSync(path.join(getHermeticDirs().runRoot, 'skill-home-'));
|
||||
const root = path.join(home, '.claude', 'skills', 'gstack');
|
||||
const stateRoot = path.join(home, '.gstack');
|
||||
try {
|
||||
// Some generated workflows keep snapshots under ~/.gstack independently
|
||||
// of GSTACK_HOME. This is owned by the same disposable HOME and cleanup.
|
||||
fs.mkdirSync(stateRoot);
|
||||
fs.mkdirSync(path.dirname(root), { recursive: true });
|
||||
fs.symlinkSync(path.resolve(import.meta.dir, '..', '..'), root, 'dir');
|
||||
// Setup exposes both the gstack runtime checkout and flattened skill
|
||||
@@ -37,7 +41,7 @@ export function hermeticSkillRuntime(): { home: string; root: string } {
|
||||
fs.rmSync(home, { recursive: true, force: true });
|
||||
throw error;
|
||||
}
|
||||
cachedRuntime = { home, root };
|
||||
cachedRuntime = { home, root, stateRoot };
|
||||
return cachedRuntime;
|
||||
}
|
||||
|
||||
@@ -55,7 +59,7 @@ function browserCache(env: Record<string, string>, originalHome: string): string
|
||||
export function withHermeticSkillRuntime(
|
||||
env: Record<string, string>,
|
||||
operatorEnv: NodeJS.ProcessEnv = process.env,
|
||||
): { env: Record<string, string>; root: string } {
|
||||
): { env: Record<string, string>; root: string; stateRoot: string } {
|
||||
const originalHome = env.HOME || os.homedir();
|
||||
const runtime = hermeticSkillRuntime();
|
||||
// An explicit empty per-test value resets Codex to its original HOME default.
|
||||
@@ -64,6 +68,7 @@ export function withHermeticSkillRuntime(
|
||||
: operatorEnv.CODEX_HOME || path.join(originalHome, '.codex');
|
||||
return {
|
||||
root: runtime.root,
|
||||
stateRoot: runtime.stateRoot,
|
||||
env: {
|
||||
...env,
|
||||
HOME: runtime.home,
|
||||
|
||||
@@ -73,12 +73,15 @@ export function recordFilePermission(input: string, file: string, cwd: string, c
|
||||
/** A long diff can crop its path header; the native access choice repeats the directory. */
|
||||
function croppedEditTarget(screen: string): string | undefined {
|
||||
const text = screen.replace(/\r+\n?/g, '\n');
|
||||
// Require actual diff rows above the native footer, never a quoted AUQ.
|
||||
if (!/^\s*\d+\s+[ +\-]?/.test(text) || /[☐□]|^\s*(?:>|`{3}|~{3})/m.test(text)) return undefined;
|
||||
// Cropping may begin inside a wrapped added/deleted diff row (five-space gutter).
|
||||
// Still require numbered rows below and the full native footer; never a quoted AUQ.
|
||||
if (!/^(?:\s*\d+\s+[ +\-]?| {5}[+\-])/.test(text) || /[☐□]|^\s*(?:>|`{3}|~{3})/m.test(text)) return undefined;
|
||||
const prompt = [...text.matchAll(/^ {0,3}Do you want to make this edit to ([^\n?\/\\]+)\?[ \t]*\n([\s\S]*)$/gm)].at(-1);
|
||||
if (!prompt || (text.slice(0, prompt.index).match(/^\s*\d+\s+/gm)?.length ?? 0) < 2) return undefined;
|
||||
// The unselected option supplies path identity only. Input remains one-time Yes.
|
||||
const choices = /^ {0,3}❯[ \t]*1\.[ \t]*Yes[ \t]*\n\s*2\.[ \t]*Yes,\s+and\s+switch\s+to\s+accept\s+edits\s+\(auto-approve\s+file\s+edits\s+and\s+common\s+file\s+commands\)\s+for\s+this\s+session;\s+Yes,\s+and\s+always\s+allow\s+access\s+to\s+([^\r\n]+?)\s+for\s+this\s+session(?:\s*\(shift\+tab\))?\s*\n\s*3\.[ \t]*No[ \t]*\n\s*Esc to cancel [·•] Tab to amend\s*$/.exec(prompt[2]!);
|
||||
// A redraw can leave this exact keyboard-hint tail on the unselected No row.
|
||||
// It does not change the selected one-time Yes or authorize another action.
|
||||
const choices = /^ {0,3}❯[ \t]*1\.[ \t]*Yes[ \t]*\n\s*2\.[ \t]*Yes,\s+and\s+switch\s+to\s+accept\s+edits\s+\(auto-approve\s+file\s+edits\s+and\s+common\s+file\s+commands\)\s+for\s+this\s+session;\s+Yes,\s+and\s+always\s+allow\s+access\s+to\s+([^\r\n]+?)\s+for\s+this\s+session(?:\s*\(shift\+tab\))?\s*\n\s*3\.[ \t]*No(?:hift\+tab\))?[ \t]*\n\s*Esc to cancel [·•] Tab to amend\s*$/.exec(prompt[2]!);
|
||||
const directory = choices?.[1]?.trim();
|
||||
if (!directory || !path.isAbsolute(directory)) return undefined;
|
||||
return path.join(directory, prompt[1]!.trim());
|
||||
|
||||
@@ -56,6 +56,69 @@ const object = (value: unknown): value is Record<string, any> =>
|
||||
const validTimestamp = (value: unknown): value is string =>
|
||||
typeof value === 'string' && Number.isFinite(Date.parse(value));
|
||||
|
||||
/** Read one length-delimited protobuf field, rejecting malformed/ambiguous input. */
|
||||
function signatureField(bytes: Uint8Array | undefined, wanted: number): Uint8Array | undefined {
|
||||
if (!bytes) return;
|
||||
let cursor = 0;
|
||||
let result: Uint8Array | undefined;
|
||||
let seen = false;
|
||||
const integer = () => {
|
||||
let value = 0;
|
||||
for (let shift = 0; shift < 70; shift += 7) {
|
||||
if (cursor >= bytes.length) throw new Error('truncated signature');
|
||||
const byte = bytes[cursor++]!;
|
||||
value += (byte & 127) * 2 ** shift;
|
||||
if (!Number.isSafeInteger(value)) throw new Error('signature integer overflow');
|
||||
if (!(byte & 128)) return value;
|
||||
}
|
||||
throw new Error('overlong signature integer');
|
||||
};
|
||||
while (cursor < bytes.length) {
|
||||
const key = integer();
|
||||
const field = Math.floor(key / 8);
|
||||
if (field < 1 || field > 0x1fffffff) throw new Error('invalid signature field');
|
||||
if (field === wanted) {
|
||||
if (seen) throw new Error('duplicate signature field');
|
||||
seen = true;
|
||||
}
|
||||
switch (key % 8) {
|
||||
case 0: integer(); break;
|
||||
case 1: cursor += 8; break;
|
||||
case 2: {
|
||||
const length = integer();
|
||||
if (length > bytes.length - cursor) throw new Error('truncated signature field');
|
||||
if (field === wanted) result = bytes.subarray(cursor, cursor + length);
|
||||
cursor += length;
|
||||
break;
|
||||
}
|
||||
case 5: cursor += 4; break;
|
||||
default: throw new Error('unsupported signature wire type');
|
||||
}
|
||||
if (cursor > bytes.length) throw new Error('truncated signature field');
|
||||
}
|
||||
return result;
|
||||
}
|
||||
|
||||
/**
|
||||
* Claude's public narration renderer classifies signature fields 2→1→8 as
|
||||
* block_kind="narration": summaries of inter-tool prose, not private reasoning.
|
||||
* Match that metadata only in this already-owned native transcript. This is
|
||||
* classification, not cryptographic signature verification. Never read the
|
||||
* thinking text of an untagged, unknown, malformed or legacy block.
|
||||
*/
|
||||
function publicNarrationText(block: Record<string, any>): string | undefined {
|
||||
if (block.type !== 'thinking' || typeof block.signature !== 'string' ||
|
||||
block.signature.length > 64 * 1024 || !/^[A-Za-z0-9+/]+={0,2}$/.test(block.signature)) return;
|
||||
try {
|
||||
const bytes = Buffer.from(block.signature, 'base64');
|
||||
const canonical = bytes.toString('base64');
|
||||
if (block.signature !== canonical && block.signature !== canonical.replace(/=+$/, '')) return;
|
||||
const tag = signatureField(signatureField(signatureField(bytes, 2), 1), 8);
|
||||
if (!tag || Buffer.from(tag).toString('utf8') !== 'narration') return;
|
||||
return typeof block.thinking === 'string' && block.thinking.trim() ? block.thinking : undefined;
|
||||
} catch { return; }
|
||||
}
|
||||
|
||||
function validQuestions(value: unknown): value is NativePlanQuestion[] {
|
||||
return Array.isArray(value) && value.length > 0 && value.every(q =>
|
||||
object(q) && typeof q.header === 'string' && typeof q.question === 'string' && q.question.trim() &&
|
||||
@@ -116,9 +179,10 @@ export function readPlanCountTranscript(configDir: string, cwd: string,
|
||||
}
|
||||
}
|
||||
|
||||
if (record.message.role === 'assistant' && block.type === 'text' &&
|
||||
typeof block.text === 'string' && block.text.trim() && validTimestamp(record.timestamp)) {
|
||||
assistantMessages.push({ sessionId: record.sessionId, text: block.text, timestamp: record.timestamp });
|
||||
if (record.message.role === 'assistant' && validTimestamp(record.timestamp)) {
|
||||
const text = block.type === 'text' && typeof block.text === 'string' && block.text.trim()
|
||||
? block.text : publicNarrationText(block);
|
||||
if (text) assistantMessages.push({ sessionId: record.sessionId, text, timestamp: record.timestamp });
|
||||
}
|
||||
if (record.message.role === 'assistant' && block.type === 'tool_use' && block.name === 'ExitPlanMode' &&
|
||||
typeof block.id === 'string' && validTimestamp(record.timestamp)) {
|
||||
|
||||
@@ -139,7 +139,7 @@ export const E2E_TOUCHFILES: Record<string, string[]> = {
|
||||
// Each one tests behavior the SDK harness can't observe (rendered TTY,
|
||||
// numbered-option lists, multi-phase ordering, idempotency state echo).
|
||||
'auq-format-gate': ['plan-ceo-review/**', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble/generate-completeness-section.ts', 'scripts/resolvers/preamble.ts', 'test/helpers/auq-sdk-capture.ts', 'test/helpers/session-runner.ts', 'test/helpers/llm-judge.ts', 'test/skill-e2e-ask-user-question-format-compliance.test.ts'],
|
||||
'plan-ceo-mode-routing': ['plan-ceo-review/**', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble.ts', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/ceo-mode-option.ts', 'test/ceo-mode-option.test.ts', 'test/pty-option-selection.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/skill-e2e-plan-ceo-mode-routing.test.ts', 'test/ceo-mode-posture-native.test.ts', 'test/fixtures/ceo-hold-posture-l.json', 'test/ceo-mode-labels-native.test.ts', 'test/fixtures/ceo-mode-labels-l.json', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/ceo-mode-prerequisite.test.ts', 'test/fixtures/ceo-mode-prerequisite-o-calls.json', 'test/fixtures/ceo-mode-prerequisite-q-call.json', 'test/plan-count-preview-footer.test.ts', 'test/fixtures/ceo-preview-u-call.json', 'test/fixtures/ceo-preview-u-screen.txt', 'test/ceo-posture-packet.test.ts', 'test/fixtures/design-preview-v-screen.txt', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/fixtures/ceo-mode-preview-aa-screen.txt', 'test/fixtures/ceo-count-mode-preview-aa-screen.txt', 'test/ceo-expansion-auq.test.ts', 'test/fixtures/ceo-expansion-auq-ac.json', 'test/helpers/plan-count-pending-question.ts', 'test/autoplan-pending-question.test.ts', 'test/plan-pending-question-pty.test.ts', 'test/ceo-barless-submit.test.ts', 'test/fixtures/ceo-barless-submit-ac.json', 'test/pending-question-completion.test.ts', 'test/fixtures/pending-question-completion-ad.json', 'test/ceo-mode-posture-ad.test.ts', 'test/fixtures/ceo-mode-posture-ad.json'],
|
||||
'plan-ceo-mode-routing': ['plan-ceo-review/**', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble.ts', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/ceo-mode-option.ts', 'test/ceo-mode-option.test.ts', 'test/pty-option-selection.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/autoplan-public-narration.test.ts', 'test/fixtures/autoplan-public-narration-ad.json', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/skill-e2e-plan-ceo-mode-routing.test.ts', 'test/ceo-mode-posture-native.test.ts', 'test/fixtures/ceo-hold-posture-l.json', 'test/ceo-mode-labels-native.test.ts', 'test/fixtures/ceo-mode-labels-l.json', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/ceo-mode-prerequisite.test.ts', 'test/fixtures/ceo-mode-prerequisite-o-calls.json', 'test/fixtures/ceo-mode-prerequisite-q-call.json', 'test/plan-count-preview-footer.test.ts', 'test/fixtures/ceo-preview-u-call.json', 'test/fixtures/ceo-preview-u-screen.txt', 'test/ceo-posture-packet.test.ts', 'test/fixtures/design-preview-v-screen.txt', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/fixtures/ceo-mode-preview-aa-screen.txt', 'test/fixtures/ceo-count-mode-preview-aa-screen.txt', 'test/ceo-expansion-auq.test.ts', 'test/fixtures/ceo-expansion-auq-ac.json', 'test/helpers/plan-count-pending-question.ts', 'test/autoplan-pending-question.test.ts', 'test/plan-pending-question-pty.test.ts', 'test/ceo-barless-submit.test.ts', 'test/fixtures/ceo-barless-submit-ac.json', 'test/pending-question-completion.test.ts', 'test/fixtures/pending-question-completion-ad.json', 'test/ceo-mode-posture-ad.test.ts', 'test/fixtures/ceo-mode-posture-ad.json', 'test/ceo-mode-full-ad.test.ts', 'test/fixtures/ceo-mode-full-ad.json'],
|
||||
'plan-design-with-ui-scope': ['plan-design-review/**', 'test/fixtures/plans/ui-heavy-feature.md', 'test/helpers/claude-pty-runner.ts', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/skill-e2e-plan-design-with-ui.test.ts', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt'],
|
||||
'ship-idempotency-pty': ['ship/**', 'bin/gstack-next-version', 'bin/gstack-version-bump', 'scripts/resolvers/sections.ts', 'lib/worktree.ts', 'test/helpers/claude-pty-runner.ts', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/skill-e2e-ship-idempotency.test.ts', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt'],
|
||||
'tpa-present': ['scripts/resolvers/third-party-actions.ts', 'ship/SKILL.md.tmpl', 'ship/sections/apple-release.md.tmpl', 'scripts/gen-skill-docs.ts', 'test/helpers/session-runner.ts', 'test/skill-e2e-third-party-actions.test.ts', 'test/helpers/third-party-actions.ts'],
|
||||
@@ -154,16 +154,16 @@ export const E2E_TOUCHFILES: Record<string, string[]> = {
|
||||
// the selector sets GSTACK_CARVE_SKILL=<name> to scope cost to the changed
|
||||
// skill (D-CODEX A). Touching the registry/helper or sections.ts runs all.
|
||||
'carve-section-loading': ['scripts/resolvers/composition.ts', 'test/autoplan-review-discovery.test.ts', 'test/autoplan-phase-order.test.ts', 'design-html/**', 'design-shotgun/**', 'qa/**', 'browse/**', 'retro/**', 'autoplan/**', 'spec/**', 'setup-gbrain/**', 'review/**', 'codex/**', 'land-and-deploy/**', 'plan-eng-review/**', 'plan-design-review/**', 'plan-devex-review/**', 'office-hours/**', 'document-release/**', 'design-consultation/**', 'cso/**', 'test/helpers/carve-guards.ts', 'scripts/resolvers/sections.ts', 'scripts/resolvers/redact-doc.ts', 'scripts/gen-skill-docs.ts', 'test/helpers/auq-sdk-capture.ts', 'test/helpers/session-runner.ts', 'test/session-runner-tools.test.ts', 'test/carve-section-loading.test.ts', 'bin/gstack-autoplan-snapshot.ts', 'test/autoplan-snapshot.test.ts', 'test/autoplan-init.test.ts', 'test/autoplan-obligations.test.ts', 'test/fixtures/autoplan/t-ceo-omitted-obligations.json', 'test/fixtures/autoplan/u-ceo-original-loss.json', 'test/fixtures/autoplan/v-ceo-dangling-references.json'],
|
||||
'autoplan-chain-pty': ['scripts/resolvers/composition.ts', 'test/autoplan-review-discovery.test.ts', 'test/autoplan-phase-order.test.ts', 'autoplan/**', 'plan-ceo-review/**', 'plan-design-review/**', 'plan-eng-review/**', 'plan-devex-review/**', 'test/fixtures/plans/autoplan-dashboard.md', 'test/autoplan-chain-fixture.test.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'bin/gstack-config', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/autoplan-setup-question.ts', 'test/autoplan-setup-question.test.ts', 'test/helpers/autoplan-phase-observer.ts', 'test/autoplan-phase-observer.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/skill-e2e-autoplan-chain.test.ts', 'bin/gstack-autoplan-snapshot.ts', 'test/autoplan-snapshot.test.ts', 'test/autoplan-init.test.ts', 'test/autoplan-obligations.test.ts', 'test/fixtures/autoplan/t-ceo-omitted-obligations.json', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/helpers/eval-budgets.ts', 'test/autoplan-eval-budget.test.ts', 'test/eval-budgets-policy.test.ts', 'test/eval-detach-timeout-floor.test.ts', 'scripts/test-paid-shards.ts', '.github/workflows/evals-periodic.yml', 'test/fixtures/autoplan-routing-n-screen.txt', 'test/autoplan-routing-o.test.ts', 'test/fixtures/autoplan-routing-o-screen.txt', 'test/autoplan-setup-packet-o.test.ts', 'test/fixtures/autoplan-setup-packet-o-screen.txt', 'test/fixtures/autoplan-setup-packet-o-call.json', 'test/fixtures/autoplan/u-ceo-original-loss.json', 'test/fixtures/autoplan/v-ceo-dangling-references.json', 'test/fixtures/autoplan-setup-z-packet.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/helpers/autoplan-method-read-audit.ts', 'test/autoplan-method-read-audit.test.ts', 'test/fixtures/autoplan-method-read-aa-events.json', 'test/autoplan-routing-manual-skills.test.ts', 'test/fixtures/autoplan-routing-manual-skills-ac.json', 'test/helpers/plan-count-pending-question.ts', 'test/autoplan-pending-question.test.ts', 'test/plan-pending-question-pty.test.ts', 'test/pending-question-completion.test.ts', 'test/fixtures/pending-question-completion-ad.json'],
|
||||
'autoplan-chain-pty': ['scripts/resolvers/composition.ts', 'test/autoplan-review-discovery.test.ts', 'test/autoplan-phase-order.test.ts', 'autoplan/**', 'plan-ceo-review/**', 'plan-design-review/**', 'plan-eng-review/**', 'plan-devex-review/**', 'test/fixtures/plans/autoplan-dashboard.md', 'test/autoplan-chain-fixture.test.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'bin/gstack-config', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/autoplan-setup-question.ts', 'test/autoplan-setup-question.test.ts', 'test/helpers/autoplan-phase-observer.ts', 'test/autoplan-phase-observer.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/autoplan-public-narration.test.ts', 'test/fixtures/autoplan-public-narration-ad.json', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/skill-e2e-autoplan-chain.test.ts', 'bin/gstack-autoplan-snapshot.ts', 'test/autoplan-snapshot.test.ts', 'test/autoplan-init.test.ts', 'test/autoplan-obligations.test.ts', 'test/fixtures/autoplan/t-ceo-omitted-obligations.json', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/helpers/eval-budgets.ts', 'test/autoplan-eval-budget.test.ts', 'test/eval-budgets-policy.test.ts', 'test/eval-detach-timeout-floor.test.ts', 'scripts/test-paid-shards.ts', '.github/workflows/evals-periodic.yml', 'test/fixtures/autoplan-routing-n-screen.txt', 'test/autoplan-routing-o.test.ts', 'test/fixtures/autoplan-routing-o-screen.txt', 'test/autoplan-setup-packet-o.test.ts', 'test/fixtures/autoplan-setup-packet-o-screen.txt', 'test/fixtures/autoplan-setup-packet-o-call.json', 'test/fixtures/autoplan/u-ceo-original-loss.json', 'test/fixtures/autoplan/v-ceo-dangling-references.json', 'test/fixtures/autoplan-setup-z-packet.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/helpers/autoplan-method-read-audit.ts', 'test/autoplan-method-read-audit.test.ts', 'test/fixtures/autoplan-method-read-aa-events.json', 'test/autoplan-routing-manual-skills.test.ts', 'test/fixtures/autoplan-routing-manual-skills-ac.json', 'test/helpers/plan-count-pending-question.ts', 'test/autoplan-pending-question.test.ts', 'test/plan-pending-question-pty.test.ts', 'test/pending-question-completion.test.ts', 'test/fixtures/pending-question-completion-ad.json'],
|
||||
|
||||
// Per-finding AskUserQuestion count + review-report-at-bottom assertion.
|
||||
// Each test drives its skill end-to-end; touchfiles include preamble +
|
||||
// completion-status resolvers because they affect question cadence and
|
||||
// terminal output (the regression surface this test catches).
|
||||
'plan-ceo-finding-count': ['bin/gstack-config', 'bin/gstack-skill-start', 'bin/gstack-skill-end', 'plan-ceo-review/**', 'test/skill-ceo-section-ordering.test.ts', 'scripts/resolvers/preamble.ts', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble/generate-completion-status.ts', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/claude-pty-runner.unit.test.ts', 'test/plan-count-completion.test.ts', 'test/plan-count-dx-handoff.test.ts', 'test/fixtures/devex-handoff-n-call.json', 'test/helpers/ceo-completion-handoff.ts', 'test/ceo-completion-handoff.test.ts', 'test/ceo-count-s-terminals.test.ts', 'test/fixtures/ceo-count-s-paired.json', 'test/fixtures/ceo-completion-handoff-calls.json', 'test/fixtures/ceo-completion-handoff-j-calls.json', 'test/fixtures/ceo-completion-handoff-k-calls.json', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/helpers/plan-count-pending-exit.ts', 'test/plan-count-pending-exit.test.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/helpers/eval-store.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-timeout.test.ts', 'test/plan-count-navigation-r.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/skill-e2e-plan-ceo-finding-count.test.ts', 'test/ceo-completion-handoff-l.test.ts', 'test/fixtures/ceo-completion-handoff-l-calls.json', 'test/ceo-completion-handoff-m.test.ts', 'test/fixtures/ceo-completion-handoff-m-call.json', 'test/fixtures/devex-review-l-calls.json', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/fixtures/ceo-handoff-n-calls.json', 'test/ceo-completion-handoff-o.test.ts', 'test/fixtures/ceo-completion-handoff-o-call.json', 'test/plan-count-empty-review.test.ts', 'test/fixtures/ceo-count-s-distinct.json', 'test/fixtures/plan-count-design-questionless-report.md', 'test/plan-count-dx-handoff-o.test.ts', 'test/fixtures/devex-handoff-o-call.json', 'test/helpers/ceo-approach-pick.ts', 'test/ceo-approach-pick.test.ts', 'test/fixtures/ceo-approach-q-call.json', 'test/fixtures/ceo-approach-r-call.json', 'test/fixtures/ceo-approach-r-distinct-call.json', 'test/fixtures/ceo-approach-q-paired-call.json', 'test/fixtures/ceo-completion-handoff-q-call.json', 'test/fixtures/ceo-completion-handoff-r-calls.json', 'test/fixtures/ceo-completion-handoff-t-call.json', 'test/helpers/plan-count-file-permission.ts', 'test/plan-count-file-permission.test.ts', 'test/fixtures/plan-count-edit-permission-t.json', 'test/plan-count-preview-footer.test.ts', 'test/fixtures/ceo-preview-u-call.json', 'test/fixtures/ceo-preview-u-screen.txt', 'test/fixtures/ceo-completion-handoff-u-call.json', 'test/fixtures/ceo-completion-handoff-v-call.json', 'test/fixtures/design-preview-v-screen.txt', 'test/plan-count-owned-permission.test.ts', 'test/fixtures/plan-count-owned-permission-v.json', 'test/fixtures/ceo-questionless-w-native.json', 'test/plan-count-ceo-body-finding.test.ts', 'test/fixtures/ceo-count-w-paired.json', 'test/fixtures/ceo-completion-handoff-w-call.json', 'test/fixtures/ceo-approach-y-call.json', 'test/fixtures/ceo-approach-y-screen.txt', 'test/ceo-handoff-y.test.ts', 'test/fixtures/ceo-handoff-y-call.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/fixtures/ceo-handoff-z-call.json', 'test/fixtures/ceo-approach-aa-call.json', 'test/review-handoffs-aa.test.ts', 'test/fixtures/review-handoff-aa-ceo.json', 'test/fixtures/review-handoff-aa-dx.json', 'test/helpers/ceo-mode-option.ts', 'test/ceo-count-mode.test.ts', 'test/fixtures/ceo-count-mode-ab-call.json', 'test/ceo-count-ac.test.ts', 'test/fixtures/ceo-count-ac-calls.json', 'test/fixtures/ceo-count-ac-later-calls.json', 'test/plan-count-permission-ac.test.ts', 'test/fixtures/plan-count-permission-ac.json'],
|
||||
'plan-eng-finding-count': ['bin/gstack-config', 'bin/gstack-skill-start', 'bin/gstack-skill-end', 'plan-eng-review/**', 'scripts/resolvers/preamble.ts', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble/generate-completion-status.ts', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/claude-pty-runner.unit.test.ts', 'test/plan-count-completion.test.ts', 'test/plan-count-dx-handoff.test.ts', 'test/fixtures/devex-handoff-n-call.json', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/helpers/plan-count-pending-exit.ts', 'test/plan-count-pending-exit.test.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/helpers/eval-store.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-timeout.test.ts', 'test/plan-count-navigation-r.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/skill-e2e-plan-eng-finding-count.test.ts', 'test/fixtures/devex-review-l-calls.json', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/plan-count-empty-review.test.ts', 'test/fixtures/ceo-count-s-distinct.json', 'test/fixtures/plan-count-design-questionless-report.md', 'test/plan-count-dx-handoff-o.test.ts', 'test/fixtures/devex-handoff-o-call.json', 'test/eng-devex-s-count.test.ts', 'test/fixtures/eng-devex-s-first-calls.json', 'test/fixtures/eng-devex-s-retry-calls.json', 'test/helpers/plan-count-file-permission.ts', 'test/plan-count-file-permission.test.ts', 'test/fixtures/plan-count-edit-permission-t.json', 'test/eng-first-review-t.test.ts', 'test/fixtures/eng-batching-t-calls.json', 'test/plan-count-preview-footer.test.ts', 'test/fixtures/ceo-preview-u-call.json', 'test/fixtures/ceo-preview-u-screen.txt', 'test/fixtures/design-preview-v-screen.txt', 'test/plan-count-owned-permission.test.ts', 'test/fixtures/plan-count-owned-permission-v.json', 'test/fixtures/ceo-questionless-w-native.json', 'test/eng-scope-y.test.ts', 'test/fixtures/eng-scope-y-calls.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/eng-binding-z.test.ts', 'test/fixtures/eng-binding-z-calls.json', 'test/eng-binding-retry-z.test.ts', 'test/fixtures/eng-binding-retry-z-calls.json', 'test/plan-count-permission-ac.test.ts', 'test/fixtures/plan-count-permission-ac.json'],
|
||||
'plan-design-finding-count': ['bin/gstack-config', 'bin/gstack-skill-start', 'bin/gstack-skill-end', 'plan-design-review/**', 'scripts/resolvers/preamble.ts', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble/generate-completion-status.ts', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/claude-pty-runner.unit.test.ts', 'test/plan-count-completion.test.ts', 'test/plan-count-dx-handoff.test.ts', 'test/fixtures/devex-handoff-n-call.json', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/helpers/plan-count-pending-exit.ts', 'test/plan-count-pending-exit.test.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/helpers/eval-store.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-timeout.test.ts', 'test/plan-count-navigation-r.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/helpers/design-count-review.ts', 'test/design-count-review.test.ts', 'test/fixtures/design-review-j-calls.json', 'test/helpers/design-count-outside.ts', 'test/design-count-outside.test.ts', 'test/skill-e2e-plan-design-finding-count.test.ts', 'test/fixtures/design-review-l-calls.json', 'test/design-completion-handoff.test.ts', 'test/fixtures/design-handoff-l-calls.json', 'test/fixtures/devex-review-l-calls.json', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/fixtures/design-review-n-calls.json', 'test/design-completion-handoff-scored.test.ts', 'test/fixtures/design-handoff-n-calls.json', 'test/fixtures/design-handoff-q-calls.json', 'test/plan-count-empty-review.test.ts', 'test/fixtures/ceo-count-s-distinct.json', 'test/fixtures/plan-count-design-questionless-report.md', 'test/plan-count-dx-handoff-o.test.ts', 'test/fixtures/devex-handoff-o-call.json', 'test/helpers/plan-count-file-permission.ts', 'test/plan-count-file-permission.test.ts', 'test/fixtures/plan-count-edit-permission-t.json', 'test/plan-count-preview-footer.test.ts', 'test/fixtures/ceo-preview-u-call.json', 'test/fixtures/ceo-preview-u-screen.txt', 'test/design-completion-handoff-u.test.ts', 'test/fixtures/design-handoff-u-calls.json', 'test/fixtures/design-preview-v-screen.txt', 'test/plan-count-owned-permission.test.ts', 'test/fixtures/plan-count-owned-permission-v.json', 'test/fixtures/ceo-questionless-w-native.json', 'test/helpers/design-artifact-question.ts', 'test/design-artifact-question.test.ts', 'test/fixtures/design-artifacts-w-calls.json', 'test/fixtures/design-outside-y-calls.json', 'test/fixtures/design-boundaries-y-calls.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/fixtures/design-gap-z-calls.json', 'test/plan-count-permission-ac.test.ts', 'test/fixtures/plan-count-permission-ac.json'],
|
||||
'plan-devex-finding-count': ['bin/gstack-config', 'bin/gstack-skill-start', 'bin/gstack-skill-end', 'plan-devex-review/**', 'scripts/resolvers/preamble.ts', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble/generate-completion-status.ts', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/claude-pty-runner.unit.test.ts', 'test/plan-count-completion.test.ts', 'test/plan-count-dx-handoff.test.ts', 'test/fixtures/devex-handoff-n-call.json', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/helpers/plan-count-pending-exit.ts', 'test/plan-count-pending-exit.test.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/helpers/eval-store.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-timeout.test.ts', 'test/plan-count-navigation-r.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/helpers/devex-count-fixture.ts', 'test/devex-count-fixture.test.ts', 'test/skill-e2e-plan-devex-finding-count.test.ts', 'test/fixtures/dx-prerequisite-r-call.json', 'test/fixtures/devex-review-l-calls.json', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/fixtures/devex-review-n-calls.json', 'test/plan-count-empty-review.test.ts', 'test/fixtures/ceo-count-s-distinct.json', 'test/fixtures/plan-count-design-questionless-report.md', 'test/devex-output-o.test.ts', 'test/fixtures/devex-review-o-calls.json', 'test/fixtures/devex-output-o-retry-call.json', 'test/devex-setup-remedy-o.test.ts', 'test/fixtures/devex-review-o-retry-calls.json', 'test/plan-count-dx-handoff-o.test.ts', 'test/fixtures/devex-handoff-o-call.json', 'test/eng-devex-s-count.test.ts', 'test/fixtures/eng-devex-s-first-calls.json', 'test/fixtures/eng-devex-s-retry-calls.json', 'test/fixtures/devex-review-t-calls.json', 'test/helpers/plan-count-file-permission.ts', 'test/plan-count-file-permission.test.ts', 'test/fixtures/plan-count-edit-permission-t.json', 'test/plan-count-preview-footer.test.ts', 'test/fixtures/ceo-preview-u-call.json', 'test/fixtures/ceo-preview-u-screen.txt', 'test/fixtures/devex-count-u-calls.json', 'test/fixtures/devex-count-u-retry-calls.json', 'test/fixtures/devex-empathy-v-calls.json', 'test/fixtures/devex-handoff-v-call.json', 'test/fixtures/design-preview-v-screen.txt', 'test/plan-count-owned-permission.test.ts', 'test/fixtures/plan-count-owned-permission-v.json', 'test/fixtures/ceo-questionless-w-native.json', 'test/fixtures/devex-count-y-calls.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/fixtures/devex-count-z-calls.json', 'test/fixtures/devex-handoff-z-call.json', 'test/review-handoffs-aa.test.ts', 'test/fixtures/review-handoff-aa-ceo.json', 'test/fixtures/review-handoff-aa-dx.json', 'test/devex-empathy-ab.test.ts', 'test/fixtures/devex-empathy-ab-calls.json', 'test/devex-ac-accounting.test.ts', 'test/fixtures/devex-ac-first-attempt-calls.json', 'test/plan-count-permission-ac.test.ts', 'test/fixtures/plan-count-permission-ac.json'],
|
||||
'plan-ceo-finding-count': ['bin/gstack-config', 'bin/gstack-skill-start', 'bin/gstack-skill-end', 'plan-ceo-review/**', 'test/skill-ceo-section-ordering.test.ts', 'scripts/resolvers/preamble.ts', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble/generate-completion-status.ts', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/claude-pty-runner.unit.test.ts', 'test/plan-count-completion.test.ts', 'test/plan-count-dx-handoff.test.ts', 'test/fixtures/devex-handoff-n-call.json', 'test/helpers/ceo-completion-handoff.ts', 'test/ceo-completion-handoff.test.ts', 'test/ceo-count-s-terminals.test.ts', 'test/fixtures/ceo-count-s-paired.json', 'test/fixtures/ceo-completion-handoff-calls.json', 'test/fixtures/ceo-completion-handoff-j-calls.json', 'test/fixtures/ceo-completion-handoff-k-calls.json', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/autoplan-public-narration.test.ts', 'test/fixtures/autoplan-public-narration-ad.json', 'test/helpers/plan-count-pending-exit.ts', 'test/plan-count-pending-exit.test.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/helpers/eval-store.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-timeout.test.ts', 'test/plan-count-navigation-r.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/skill-e2e-plan-ceo-finding-count.test.ts', 'test/ceo-completion-handoff-l.test.ts', 'test/fixtures/ceo-completion-handoff-l-calls.json', 'test/ceo-completion-handoff-m.test.ts', 'test/fixtures/ceo-completion-handoff-m-call.json', 'test/fixtures/devex-review-l-calls.json', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/fixtures/ceo-handoff-n-calls.json', 'test/ceo-completion-handoff-o.test.ts', 'test/fixtures/ceo-completion-handoff-o-call.json', 'test/plan-count-empty-review.test.ts', 'test/fixtures/ceo-count-s-distinct.json', 'test/fixtures/plan-count-design-questionless-report.md', 'test/plan-count-dx-handoff-o.test.ts', 'test/fixtures/devex-handoff-o-call.json', 'test/helpers/ceo-approach-pick.ts', 'test/ceo-approach-pick.test.ts', 'test/fixtures/ceo-approach-q-call.json', 'test/fixtures/ceo-approach-r-call.json', 'test/fixtures/ceo-approach-r-distinct-call.json', 'test/fixtures/ceo-approach-q-paired-call.json', 'test/fixtures/ceo-completion-handoff-q-call.json', 'test/fixtures/ceo-completion-handoff-r-calls.json', 'test/fixtures/ceo-completion-handoff-t-call.json', 'test/helpers/plan-count-file-permission.ts', 'test/plan-count-file-permission.test.ts', 'test/fixtures/plan-count-edit-permission-t.json', 'test/plan-count-preview-footer.test.ts', 'test/fixtures/ceo-preview-u-call.json', 'test/fixtures/ceo-preview-u-screen.txt', 'test/fixtures/ceo-completion-handoff-u-call.json', 'test/fixtures/ceo-completion-handoff-v-call.json', 'test/fixtures/design-preview-v-screen.txt', 'test/plan-count-owned-permission.test.ts', 'test/fixtures/plan-count-owned-permission-v.json', 'test/fixtures/ceo-questionless-w-native.json', 'test/plan-count-ceo-body-finding.test.ts', 'test/fixtures/ceo-count-w-paired.json', 'test/fixtures/ceo-completion-handoff-w-call.json', 'test/fixtures/ceo-approach-y-call.json', 'test/fixtures/ceo-approach-y-screen.txt', 'test/ceo-handoff-y.test.ts', 'test/fixtures/ceo-handoff-y-call.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/fixtures/ceo-handoff-z-call.json', 'test/fixtures/ceo-approach-aa-call.json', 'test/review-handoffs-aa.test.ts', 'test/fixtures/review-handoff-aa-ceo.json', 'test/fixtures/review-handoff-aa-dx.json', 'test/helpers/ceo-mode-option.ts', 'test/ceo-count-mode.test.ts', 'test/fixtures/ceo-count-mode-ab-call.json', 'test/ceo-count-ac.test.ts', 'test/fixtures/ceo-count-ac-calls.json', 'test/fixtures/ceo-count-ac-later-calls.json', 'test/plan-count-permission-ac.test.ts', 'test/fixtures/plan-count-permission-ac.json', 'test/fixtures/plan-count-permission-ad.json'],
|
||||
'plan-eng-finding-count': ['bin/gstack-config', 'bin/gstack-skill-start', 'bin/gstack-skill-end', 'plan-eng-review/**', 'scripts/resolvers/preamble.ts', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble/generate-completion-status.ts', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/claude-pty-runner.unit.test.ts', 'test/plan-count-completion.test.ts', 'test/plan-count-dx-handoff.test.ts', 'test/fixtures/devex-handoff-n-call.json', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/autoplan-public-narration.test.ts', 'test/fixtures/autoplan-public-narration-ad.json', 'test/helpers/plan-count-pending-exit.ts', 'test/plan-count-pending-exit.test.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/helpers/eval-store.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-timeout.test.ts', 'test/plan-count-navigation-r.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/skill-e2e-plan-eng-finding-count.test.ts', 'test/fixtures/devex-review-l-calls.json', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/plan-count-empty-review.test.ts', 'test/fixtures/ceo-count-s-distinct.json', 'test/fixtures/plan-count-design-questionless-report.md', 'test/plan-count-dx-handoff-o.test.ts', 'test/fixtures/devex-handoff-o-call.json', 'test/eng-devex-s-count.test.ts', 'test/fixtures/eng-devex-s-first-calls.json', 'test/fixtures/eng-devex-s-retry-calls.json', 'test/helpers/plan-count-file-permission.ts', 'test/plan-count-file-permission.test.ts', 'test/fixtures/plan-count-edit-permission-t.json', 'test/eng-first-review-t.test.ts', 'test/fixtures/eng-batching-t-calls.json', 'test/plan-count-preview-footer.test.ts', 'test/fixtures/ceo-preview-u-call.json', 'test/fixtures/ceo-preview-u-screen.txt', 'test/fixtures/design-preview-v-screen.txt', 'test/plan-count-owned-permission.test.ts', 'test/fixtures/plan-count-owned-permission-v.json', 'test/fixtures/ceo-questionless-w-native.json', 'test/eng-scope-y.test.ts', 'test/fixtures/eng-scope-y-calls.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/eng-binding-z.test.ts', 'test/fixtures/eng-binding-z-calls.json', 'test/eng-binding-retry-z.test.ts', 'test/fixtures/eng-binding-retry-z-calls.json', 'test/plan-count-permission-ac.test.ts', 'test/fixtures/plan-count-permission-ac.json', 'test/fixtures/plan-count-permission-ad.json'],
|
||||
'plan-design-finding-count': ['bin/gstack-config', 'bin/gstack-skill-start', 'bin/gstack-skill-end', 'plan-design-review/**', 'scripts/resolvers/preamble.ts', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble/generate-completion-status.ts', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/claude-pty-runner.unit.test.ts', 'test/plan-count-completion.test.ts', 'test/plan-count-dx-handoff.test.ts', 'test/fixtures/devex-handoff-n-call.json', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/autoplan-public-narration.test.ts', 'test/fixtures/autoplan-public-narration-ad.json', 'test/helpers/plan-count-pending-exit.ts', 'test/plan-count-pending-exit.test.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/helpers/eval-store.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-timeout.test.ts', 'test/plan-count-navigation-r.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/helpers/design-count-review.ts', 'test/design-count-review.test.ts', 'test/fixtures/design-review-j-calls.json', 'test/helpers/design-count-outside.ts', 'test/design-count-outside.test.ts', 'test/skill-e2e-plan-design-finding-count.test.ts', 'test/fixtures/design-review-l-calls.json', 'test/design-completion-handoff.test.ts', 'test/fixtures/design-handoff-l-calls.json', 'test/fixtures/devex-review-l-calls.json', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/fixtures/design-review-n-calls.json', 'test/design-completion-handoff-scored.test.ts', 'test/fixtures/design-handoff-n-calls.json', 'test/fixtures/design-handoff-q-calls.json', 'test/plan-count-empty-review.test.ts', 'test/fixtures/ceo-count-s-distinct.json', 'test/fixtures/plan-count-design-questionless-report.md', 'test/plan-count-dx-handoff-o.test.ts', 'test/fixtures/devex-handoff-o-call.json', 'test/helpers/plan-count-file-permission.ts', 'test/plan-count-file-permission.test.ts', 'test/fixtures/plan-count-edit-permission-t.json', 'test/plan-count-preview-footer.test.ts', 'test/fixtures/ceo-preview-u-call.json', 'test/fixtures/ceo-preview-u-screen.txt', 'test/design-completion-handoff-u.test.ts', 'test/fixtures/design-handoff-u-calls.json', 'test/fixtures/design-preview-v-screen.txt', 'test/plan-count-owned-permission.test.ts', 'test/fixtures/plan-count-owned-permission-v.json', 'test/fixtures/ceo-questionless-w-native.json', 'test/helpers/design-artifact-question.ts', 'test/design-artifact-question.test.ts', 'test/fixtures/design-artifacts-w-calls.json', 'test/fixtures/design-outside-y-calls.json', 'test/fixtures/design-boundaries-y-calls.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/fixtures/design-gap-z-calls.json', 'test/plan-count-permission-ac.test.ts', 'test/fixtures/plan-count-permission-ac.json', 'test/fixtures/plan-count-permission-ad.json'],
|
||||
'plan-devex-finding-count': ['bin/gstack-config', 'bin/gstack-skill-start', 'bin/gstack-skill-end', 'plan-devex-review/**', 'scripts/resolvers/preamble.ts', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble/generate-completion-status.ts', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/claude-pty-runner.unit.test.ts', 'test/plan-count-completion.test.ts', 'test/plan-count-dx-handoff.test.ts', 'test/fixtures/devex-handoff-n-call.json', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/autoplan-public-narration.test.ts', 'test/fixtures/autoplan-public-narration-ad.json', 'test/helpers/plan-count-pending-exit.ts', 'test/plan-count-pending-exit.test.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/helpers/eval-store.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-timeout.test.ts', 'test/plan-count-navigation-r.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/helpers/devex-count-fixture.ts', 'test/devex-count-fixture.test.ts', 'test/skill-e2e-plan-devex-finding-count.test.ts', 'test/fixtures/dx-prerequisite-r-call.json', 'test/fixtures/devex-review-l-calls.json', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/fixtures/devex-review-n-calls.json', 'test/plan-count-empty-review.test.ts', 'test/fixtures/ceo-count-s-distinct.json', 'test/fixtures/plan-count-design-questionless-report.md', 'test/devex-output-o.test.ts', 'test/fixtures/devex-review-o-calls.json', 'test/fixtures/devex-output-o-retry-call.json', 'test/devex-setup-remedy-o.test.ts', 'test/fixtures/devex-review-o-retry-calls.json', 'test/plan-count-dx-handoff-o.test.ts', 'test/fixtures/devex-handoff-o-call.json', 'test/eng-devex-s-count.test.ts', 'test/fixtures/eng-devex-s-first-calls.json', 'test/fixtures/eng-devex-s-retry-calls.json', 'test/fixtures/devex-review-t-calls.json', 'test/helpers/plan-count-file-permission.ts', 'test/plan-count-file-permission.test.ts', 'test/fixtures/plan-count-edit-permission-t.json', 'test/plan-count-preview-footer.test.ts', 'test/fixtures/ceo-preview-u-call.json', 'test/fixtures/ceo-preview-u-screen.txt', 'test/fixtures/devex-count-u-calls.json', 'test/fixtures/devex-count-u-retry-calls.json', 'test/fixtures/devex-empathy-v-calls.json', 'test/fixtures/devex-handoff-v-call.json', 'test/fixtures/design-preview-v-screen.txt', 'test/plan-count-owned-permission.test.ts', 'test/fixtures/plan-count-owned-permission-v.json', 'test/fixtures/ceo-questionless-w-native.json', 'test/fixtures/devex-count-y-calls.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/fixtures/devex-count-z-calls.json', 'test/fixtures/devex-handoff-z-call.json', 'test/review-handoffs-aa.test.ts', 'test/fixtures/review-handoff-aa-ceo.json', 'test/fixtures/review-handoff-aa-dx.json', 'test/devex-empathy-ab.test.ts', 'test/fixtures/devex-empathy-ab-calls.json', 'test/devex-ac-accounting.test.ts', 'test/fixtures/devex-ac-first-attempt-calls.json', 'test/plan-count-permission-ac.test.ts', 'test/fixtures/plan-count-permission-ac.json', 'test/fixtures/plan-count-permission-ad.json'],
|
||||
|
||||
// Gate-tier reviewCount-floor counterparts. Catch the May 2026 transcript
|
||||
// bug (model wrote a plan-mode plan and ExitPlanMode'd without firing any
|
||||
@@ -180,8 +180,8 @@ export const E2E_TOUCHFILES: Record<string, string[]> = {
|
||||
// a model fires one AUQ then batches the rest into a "## Decisions to
|
||||
// confirm" plan write. runPlanSkillFloorCheck cannot detect that shape
|
||||
// (it exits on first AUQ); runPlanSkillCounting can.
|
||||
'plan-eng-multi-finding-batching': ['bin/gstack-config', 'bin/gstack-skill-start', 'bin/gstack-skill-end', 'plan-eng-review/**', 'scripts/resolvers/preamble.ts', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble/generate-completion-status.ts', 'scripts/resolvers/review.ts', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/claude-pty-runner.unit.test.ts', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/helpers/plan-count-pending-exit.ts', 'test/plan-count-pending-exit.test.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/helpers/eval-store.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-timeout.test.ts', 'test/plan-count-navigation-r.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/fixtures/forcing-finding-seeds.ts', 'test/skill-e2e-plan-eng-multi-finding-batching.test.ts', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/eng-devex-s-count.test.ts', 'test/fixtures/eng-devex-s-first-calls.json', 'test/fixtures/eng-devex-s-retry-calls.json', 'test/helpers/plan-count-file-permission.ts', 'test/plan-count-file-permission.test.ts', 'test/fixtures/plan-count-edit-permission-t.json', 'test/eng-first-review-t.test.ts', 'test/fixtures/eng-batching-t-calls.json', 'test/plan-count-preview-footer.test.ts', 'test/fixtures/ceo-preview-u-call.json', 'test/fixtures/ceo-preview-u-screen.txt', 'test/fixtures/design-preview-v-screen.txt', 'test/plan-count-owned-permission.test.ts', 'test/fixtures/plan-count-owned-permission-v.json', 'test/fixtures/ceo-questionless-w-native.json', 'test/eng-scope-y.test.ts', 'test/fixtures/eng-scope-y-calls.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/eng-binding-z.test.ts', 'test/fixtures/eng-binding-z-calls.json', 'test/eng-binding-retry-z.test.ts', 'test/fixtures/eng-binding-retry-z-calls.json', 'test/plan-count-permission-ac.test.ts', 'test/fixtures/plan-count-permission-ac.json'],
|
||||
'plan-ceo-split-overflow': ['bin/gstack-config', 'plan-ceo-review/**', 'scripts/resolvers/preamble.ts', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'bin/gstack-question-preference', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/claude-pty-runner.unit.test.ts', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/helpers/plan-count-pending-exit.ts', 'test/plan-count-pending-exit.test.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/helpers/eval-store.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-timeout.test.ts', 'test/plan-count-navigation-r.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/fixtures/forcing-finding-seeds.ts', 'test/skill-e2e-plan-ceo-split-overflow.test.ts', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/helpers/plan-count-file-permission.ts', 'test/plan-count-file-permission.test.ts', 'test/fixtures/plan-count-edit-permission-t.json', 'test/plan-count-owned-permission.test.ts', 'test/fixtures/plan-count-owned-permission-v.json', 'test/fixtures/ceo-questionless-w-native.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/plan-count-permission-ac.test.ts', 'test/fixtures/plan-count-permission-ac.json'],
|
||||
'plan-eng-multi-finding-batching': ['bin/gstack-config', 'bin/gstack-skill-start', 'bin/gstack-skill-end', 'plan-eng-review/**', 'scripts/resolvers/preamble.ts', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'scripts/resolvers/preamble/generate-completion-status.ts', 'scripts/resolvers/review.ts', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/claude-pty-runner.unit.test.ts', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/autoplan-public-narration.test.ts', 'test/fixtures/autoplan-public-narration-ad.json', 'test/helpers/plan-count-pending-exit.ts', 'test/plan-count-pending-exit.test.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/helpers/eval-store.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-timeout.test.ts', 'test/plan-count-navigation-r.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/fixtures/forcing-finding-seeds.ts', 'test/skill-e2e-plan-eng-multi-finding-batching.test.ts', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/eng-devex-s-count.test.ts', 'test/fixtures/eng-devex-s-first-calls.json', 'test/fixtures/eng-devex-s-retry-calls.json', 'test/helpers/plan-count-file-permission.ts', 'test/plan-count-file-permission.test.ts', 'test/fixtures/plan-count-edit-permission-t.json', 'test/eng-first-review-t.test.ts', 'test/fixtures/eng-batching-t-calls.json', 'test/plan-count-preview-footer.test.ts', 'test/fixtures/ceo-preview-u-call.json', 'test/fixtures/ceo-preview-u-screen.txt', 'test/fixtures/design-preview-v-screen.txt', 'test/plan-count-owned-permission.test.ts', 'test/fixtures/plan-count-owned-permission-v.json', 'test/fixtures/ceo-questionless-w-native.json', 'test/eng-scope-y.test.ts', 'test/fixtures/eng-scope-y-calls.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/eng-binding-z.test.ts', 'test/fixtures/eng-binding-z-calls.json', 'test/eng-binding-retry-z.test.ts', 'test/fixtures/eng-binding-retry-z-calls.json', 'test/plan-count-permission-ac.test.ts', 'test/fixtures/plan-count-permission-ac.json', 'test/fixtures/plan-count-permission-ad.json'],
|
||||
'plan-ceo-split-overflow': ['bin/gstack-config', 'plan-ceo-review/**', 'scripts/resolvers/preamble.ts', 'scripts/resolvers/preamble/generate-ask-user-format.ts', 'bin/gstack-question-preference', 'test/helpers/claude-pty-runner.ts', 'test/plan-count-native-input.test.ts', 'test/helpers/pty-screen.ts', 'test/pty-screen.test.ts', 'test/pty-screen-session.test.ts', 'test/fixtures/pty-screen/**', 'test/helpers/hermetic-skill-runtime.ts', 'test/hermetic-skill-runtime.test.ts', 'test/helpers/claude-pty-runner.unit.test.ts', 'test/helpers/pty-trust-dialog.ts', 'test/pty-trust-dialog.test.ts', 'test/helpers/plan-count-transcript.ts', 'test/autoplan-public-narration.test.ts', 'test/fixtures/autoplan-public-narration-ad.json', 'test/helpers/plan-count-pending-exit.ts', 'test/plan-count-pending-exit.test.ts', 'test/plan-count-transcript.test.ts', 'test/helpers/plan-count-artifacts.ts', 'test/plan-count-artifacts.test.ts', 'test/helpers/eval-store.ts', 'test/helpers/plan-count-fixture.ts', 'test/plan-count-fixture.test.ts', 'test/plan-count-timeout.test.ts', 'test/plan-count-navigation-r.test.ts', 'test/plan-count-prerequisite-n.test.ts', 'test/fixtures/ceo-prerequisite-n-call.json', 'test/fixtures/forcing-finding-seeds.ts', 'test/skill-e2e-plan-ceo-split-overflow.test.ts', 'test/plan-count-checkbox.test.ts', 'test/fixtures/ceo-checkbox-l.screen.txt', 'test/helpers/plan-count-file-permission.ts', 'test/plan-count-file-permission.test.ts', 'test/fixtures/plan-count-edit-permission-t.json', 'test/plan-count-owned-permission.test.ts', 'test/fixtures/plan-count-owned-permission-v.json', 'test/fixtures/ceo-questionless-w-native.json', 'test/plan-count-truncated-question.test.ts', 'test/fixtures/ceo-approach-z-call.json', 'test/fixtures/ceo-approach-z-screen.txt', 'test/plan-count-permission-ac.test.ts', 'test/fixtures/plan-count-permission-ac.json', 'test/fixtures/plan-count-permission-ad.json'],
|
||||
'brain-privacy-gate': ['bin/gstack-skill-start', 'bin/gstack-skill-end', 'scripts/resolvers/preamble/generate-brain-sync-block.ts', 'scripts/resolvers/preamble.ts', 'bin/gstack-brain-sync', 'bin/gstack-artifacts-init', 'bin/gstack-config', 'test/helpers/agent-sdk-runner.ts', 'test/skill-e2e-brain-privacy-gate.test.ts'],
|
||||
|
||||
// /setup-gbrain Path 4 (Remote MCP) — happy + bad-token end-to-end via
|
||||
|
||||
@@ -46,13 +46,20 @@ import { createHash } from 'node:crypto';
|
||||
import { spawnSync } from 'node:child_process';
|
||||
const root = path.join(process.env.HOME,'.claude','skills','gstack');
|
||||
const args = process.argv.slice(2);
|
||||
const addDir = args.includes('--add-dir') ? args[args.indexOf('--add-dir')+1] : null;
|
||||
const addDirs = args.flatMap((arg,index) => arg === '--add-dir' ? [args[index+1]] : []);
|
||||
const addDir = addDirs[0] || null;
|
||||
const stateDirectory = path.join(process.env.HOME,'.gstack');
|
||||
const methodologyPath = path.join(stateDirectory,'projects','probe','autoplan-design-methodology-owned','methodology.md');
|
||||
const inside = (file,directory) => { const relative=path.relative(directory,file); return relative==='' || (!relative.startsWith('..'+path.sep) && relative!=='..' && !path.isAbsolute(relative)); };
|
||||
|
||||
const phase = spawnSync('bash',['-c','cat ~/.claude/skills/gstack/autoplan/sections/ceo-phase.md'],{timeout:5000});
|
||||
const config = spawnSync('bash',['-c','"$HOME/.claude/skills/gstack/bin/gstack-config" get codex_reviews'],{timeout:5000});
|
||||
const canonicalConfig = process.env.CLAUDE_CONFIG_DIR && spawnSync('bash',['-c','"$CLAUDE_CONFIG_DIR/skills/gstack/bin/gstack-config" get codex_reviews'],{timeout:5000});
|
||||
const codexHome = process.env.CODEX_HOME || path.join(process.env.HOME,'.codex');
|
||||
const record = {
|
||||
pid:process.pid, home:process.env.HOME, root:fs.realpathSync(root), args,
|
||||
pid:process.pid, home:process.env.HOME, root:fs.realpathSync(root), args, addDirs,
|
||||
stateDirectory, stateDirectoryExists:fs.existsSync(stateDirectory), methodologyPath,
|
||||
methodologyAllowed:addDirs.some(directory => inside(methodologyPath,directory)),
|
||||
phaseExit:phase.status, phaseHash:createHash('sha256').update(phase.stdout).digest('hex'),
|
||||
configExit:config.status, configValue:config.stdout.toString().trim(),
|
||||
canonicalConfigExit:canonicalConfig?.status, canonicalConfigValue:canonicalConfig?.stdout.toString().trim(),
|
||||
@@ -130,6 +137,16 @@ try {
|
||||
expect(result.home).not.toBe(operatorHome);
|
||||
expect(result.root).toBe(fs.realpathSync(ROOT));
|
||||
expect(result.configValue).toBe('disabled'); expect(result.runtimeAllowed).toBe(true);
|
||||
// Actual Autoplan snapshots live under generated HOME/.gstack, outside
|
||||
// the runtime checkout. Approve only that owned state tree, not HOME,
|
||||
// an inherited/explicit GSTACK_HOME, or a broad policy-setting answer.
|
||||
expect(result.methodologyAllowed, item.name + ': owned methodology Read is covered').toBe(true);
|
||||
expect(result.stateDirectoryExists).toBe(true);
|
||||
expect(result.addDirs).toEqual([path.join(result.home,'.claude','skills','gstack'),path.join(result.home,'.gstack')]);
|
||||
for (const external of [operatorHome,state,path.dirname(result.home),result.codexHome,result.browserCache])
|
||||
expect(result.addDirs).not.toContain(external);
|
||||
expect(result.args).not.toContain('--dangerously-skip-permissions');
|
||||
expect(result.args).not.toContain('--settings');
|
||||
expect(result.canonicalConfigExit).toBe(0); expect(result.canonicalConfigValue).toBe('disabled');
|
||||
for (const asset of result.discovery) {
|
||||
expect(asset.home, item.name + ': HOME discovery ' + asset.relative).toBe(fs.realpathSync(path.join(ROOT,asset.relative)));
|
||||
@@ -145,6 +162,7 @@ try {
|
||||
} else {
|
||||
expect(result.home).toBe(operatorHome); expect(result.phaseHash).toBe(originalSection);
|
||||
expect(result.configValue).toBe('stale-runtime'); expect(result.runtimeAllowed).toBe(false);
|
||||
expect(result.addDirs).toEqual([]); expect(result.methodologyAllowed).toBe(false);
|
||||
}
|
||||
if (item.name === 'explicit-config') expect(result.configDir).toBe(customConfig);
|
||||
expect(() => process.kill(result.pid,0)).toThrow();
|
||||
|
||||
@@ -63,7 +63,7 @@ describe('captured R planning navigation', () => {
|
||||
}
|
||||
});
|
||||
|
||||
test('short prerequisite labels require the current single native question and affirmative review action', () => {
|
||||
test('short prerequisite labels require the current native question and affirmative review action', () => {
|
||||
for (const change of [
|
||||
(c: NativePlanQuestionCall) => { c.questions[0]!.options[1]!.description = ''; },
|
||||
(c: NativePlanQuestionCall) => { c.questions[0]!.options[1]!.description = 'Do not proceed with standard DX POLISH review.'; },
|
||||
@@ -73,7 +73,6 @@ describe('captured R planning navigation', () => {
|
||||
(c: NativePlanQuestionCall) => { c.questions[0]!.options[1]!.description = 'Proceed with standard DX POLISH review? No, run /office-hours first.'; },
|
||||
(c: NativePlanQuestionCall) => { c.questions[0]!.options[1]!.description = 'Proceed with standard DX POLISH review. Accept the security risk.'; },
|
||||
(c: NativePlanQuestionCall) => { c.questions[0]!.options[1]!.description = 'Plan scope is already precise. Proceed with standard DX POLISH review if the tests pass.'; },
|
||||
(c: NativePlanQuestionCall) => { c.questions[0]!.options[1]!.label = 'Skip'; },
|
||||
(c: NativePlanQuestionCall) => { c.questions[0]!.options.push({ label: 'Accept this security finding' }); },
|
||||
(c: NativePlanQuestionCall) => { c.questions[0]!.multiSelect = true; },
|
||||
(c: NativePlanQuestionCall) => { c.questions.push(structuredClone(c.questions[0]!)); },
|
||||
@@ -99,6 +98,11 @@ describe('captured R planning navigation', () => {
|
||||
expect(planCountPrerequisitePick(reordered.routing, reordered.active)).toBe(1);
|
||||
});
|
||||
|
||||
test('a bare short label uses its bound unconditional review meaning', () => {
|
||||
const call = pending(prerequisite as NativePlanQuestionCall);call.questions[0]!.options[1]!.label='Skip';
|
||||
const {active,routing}=frame(call);expect(planCountPrerequisitePick(routing,active)).toBe(2);
|
||||
});
|
||||
|
||||
test('short labels allow only a benign plan-scope premise plus the unconditional review action', () => {
|
||||
for (const description of ['Proceed with standard review.', 'Proceed with standard DX review', 'The plan is clear. Proceed with standard DX POLISH review.', 'Plan scope is already precise. Proceed with standard DX POLISH review.']) {
|
||||
const call = pending(prerequisite as NativePlanQuestionCall);
|
||||
|
||||
@@ -3,6 +3,7 @@ import * as fs from 'node:fs';
|
||||
import * as os from 'node:os';
|
||||
import * as path from 'node:path';
|
||||
import captured from './fixtures/plan-count-permission-ac.json';
|
||||
import capturedAd from './fixtures/plan-count-permission-ad.json';
|
||||
import { classifyPlanCountFrame, createPlanCountPermissionGuard } from './helpers/claude-pty-runner';
|
||||
import { recordFilePermission, currentFilePermissionEpoch, currentFilePermissionBinding } from './helpers/plan-count-file-permission';
|
||||
import { E2E_TOUCHFILES, selectTests } from './helpers/touchfiles';
|
||||
@@ -147,3 +148,89 @@ test('a later exact owned binding wins over an earlier same-basename block', ()
|
||||
expect(createPlanCountPermissionGuard()(otherScreen, '', unrelated)).toBe('grant');
|
||||
} finally { f.close(); }
|
||||
});
|
||||
|
||||
// Exact current screens plus content-free native identity from the full AD run.
|
||||
// The replay projections do not assert these pending writes ever completed.
|
||||
const cases = capturedAd.rows.map(row => ({
|
||||
p: row, screen: row.screen, binding: {expected: row.state.expected, state: row.state},
|
||||
observation: {transcript: {status: row.transcriptStatus, calls: [],
|
||||
assistantMessages: row.transcriptSessions.map(sessionId => ({sessionId}))}},
|
||||
}));
|
||||
function adEpoch(c: any, screen = c.screen, state = c.binding.state, delta: any = {}) {
|
||||
const dir = fs.mkdtempSync(path.join(os.tmpdir(), 'permission-crop-replay-'));
|
||||
const file = path.join(dir, 'state.json');
|
||||
try {
|
||||
fs.writeFileSync(file, JSON.stringify(state));
|
||||
return currentFilePermissionEpoch(file, delta.expected ?? c.binding.expected, delta.cwd ?? c.p.cwd,
|
||||
delta.config ?? c.p.config, delta.startedAt ?? c.p.startUnix * 1000,
|
||||
delta.transcript ?? c.observation.transcript, screen);
|
||||
} finally { fs.rmSync(dir, {recursive:true, force:true}); }
|
||||
}
|
||||
for (const c of cases) {
|
||||
test(`actual AD captured current permission ${c.p.pid} returns its exact pending epoch`, () => {
|
||||
expect(classifyPlanCountFrame(c.screen)).toBe('permission');
|
||||
expect(adEpoch(c)).toEqual({pendingId:c.binding.state.pendingId, completedId:c.binding.state.completedId,
|
||||
completedIds:c.binding.state.completedIds});
|
||||
const guard = createPlanCountPermissionGuard();
|
||||
expect(guard(c.screen, '', adEpoch(c))).toBe('grant');
|
||||
expect(guard(c.screen, '', adEpoch(c))).toBe('handled');
|
||||
});
|
||||
test(`AD isolating the rejected rendering guard ${c.p.pid} preserves native identity`, () => {
|
||||
// These are explicitly normalized controls; the actual captured screen is unchanged above.
|
||||
const normalized = c.p.pid === 1332470 ? c.screen.replace('3. Nohift+tab)', '3. No') : c.screen.replace(/^ \+/, ' 99 +');
|
||||
expect(normalized).not.toBe(c.screen);
|
||||
expect(adEpoch(c, normalized)?.pendingId).toBe(c.binding.state.pendingId);
|
||||
});
|
||||
}
|
||||
for (const c of cases) {
|
||||
test(`AD crop ${c.p.pid} rejects foreign, quoted, incomplete and policy-changing menus`, () => {
|
||||
const directory = path.dirname(c.binding.expected);
|
||||
const changes: [string, string][] = [
|
||||
['foreign directory', c.screen.replace(directory, path.join(directory, 'foreign'))],
|
||||
['split directory', c.screen.replace(directory, directory + '\n/foreign')],
|
||||
['quoted', '> Example:\n' + c.screen],
|
||||
['AUQ', '☐ Review\n' + c.screen],
|
||||
['code fence', '```\n' + c.screen],
|
||||
['missing footer', c.screen.replace('Esc to cancel · Tab to amend', '')],
|
||||
['policy on selected Yes', c.screen.replace('❯ 1. Yes', '❯ 1. Yes, always allow')],
|
||||
['selected No', c.screen.replace('❯ 1. Yes', ' 1. Yes').replace(' 3. No', ' ❯ 3. No')],
|
||||
['arbitrary No suffix', c.screen.replace(/3\. No(?:hift\+tab\))?/, '3. No; run another command')],
|
||||
['another hint', c.screen.replace(/3\. No(?:hift\+tab\))?/, '3. No(shift+enter)')],
|
||||
['foreign option action', c.screen.replace(/always\s+allow\s+access\s+to/, 'delete files from')],
|
||||
['unrecognized cropped prose', c.screen.replace(/^.*\n/, 'arbitrary text\n')],
|
||||
];
|
||||
for (const [name, screen] of changes) {
|
||||
expect(screen, name).not.toBe(c.screen);
|
||||
expect(adEpoch(c, screen), name).not.toBeTruthy();
|
||||
expect(createPlanCountPermissionGuard()(screen, '', adEpoch(c, screen)), name).not.toBe('grant');
|
||||
}
|
||||
});
|
||||
test(`AD crop ${c.p.pid} leaves unrelated-basename permission policy unchanged`, () => {
|
||||
const screen = c.screen.replace(path.basename(c.binding.expected), 'OTHER.md');
|
||||
expect(adEpoch(c, screen)).toBeUndefined();
|
||||
// This is intentionally the existing caller policy for unrelated fixture permissions.
|
||||
expect(createPlanCountPermissionGuard()(screen, '', adEpoch(c, screen))).toBe('grant');
|
||||
});
|
||||
test(`AD crop ${c.p.pid} cannot replace missing, stale, completed or foreign native identity`, () => {
|
||||
const original = c.binding.state;
|
||||
for (const [name, state, delta] of [
|
||||
['foreign session', {...original, sessionId:'foreign'}, {}],
|
||||
['wrong native transcript', {...original, transcriptPath:path.join(c.p.config, 'projects', 'foreign', 'other.jsonl')}, {}],
|
||||
['completed request', {...original, completedId:original.pendingId}, {}],
|
||||
['no pending request', {...original, pendingId:null}, {}],
|
||||
['unseen pending request', {...original, pendingId:original.sessionId + ':other'}, {}],
|
||||
['stale timestamp', {...original, timestamp:new Date(c.p.startUnix * 1000 - 1).toISOString()}, {}],
|
||||
['future timestamp', {...original, timestamp:new Date(Date.now() + 60_000).toISOString()}, {}],
|
||||
['mixed sessions', original, {transcript:{status:'ready', calls:[], assistantMessages:[{sessionId:original.sessionId},{sessionId:'foreign'}]}}],
|
||||
['unready transcript', original, {transcript:{...c.observation.transcript,status:'unavailable'}}],
|
||||
] as const) {
|
||||
expect(adEpoch(c, c.screen, state, delta), name).toBeNull();
|
||||
}
|
||||
});
|
||||
}
|
||||
|
||||
test('AD crop fixture selects the exact existing permission regression callers', () => {
|
||||
expect(selectTests(['test/fixtures/plan-count-permission-ad.json'], E2E_TOUCHFILES).selected.sort()).toEqual(
|
||||
selectTests(['test/fixtures/plan-count-permission-ac.json'], E2E_TOUCHFILES).selected.sort());
|
||||
expect(selectTests(['test/fixtures/plan-count-permission-ad.json'], E2E_TOUCHFILES).selected).toContain('plan-ceo-finding-count');
|
||||
});
|
||||
Reference in new issue
Block a user