CalvinBackup/gstack
1
0
Fork 0
mirror of https://github.com/garrytan/gstack.git synced 2026-09-17 18:32:19 +02:00
Code Issues Packages Projects Releases Wiki Activity

v1.86.0.0 feat: route outside reviews by harness (#2850)

Browse Source 
* feat: add a restricted and supervised Claude Code runner

Preserve configured authentication and models while enforcing tool access, strict completion JSON, bounded output and process cleanup. Cover argv, failure handling, session metadata and Windows process containment.

* feat: route outside reviews by harness and migrate wrapper installs

Use Claude Code from Codex and Codex from other supported hosts, with shared invocation rendering, positive gate validation and per-phase provenance. Rename /claude to /claude-code, repair managed shared and copied installations safely, and generate native Kiro skills. Add installed-workflow, failure-injection and live cross-harness regression coverage.

* test: recognize CEO mode labels without terminal spacing

The paid workflow rendered SCOPEEXPANSION at option 4, but its driver required a literal space. Match the leading mode title without cursor-spacing artifacts and ignore adjacent preview text. Preserve missing-target failures and downstream posture assertions.

* test: isolate plan-count fixtures before starting review workflows

Seed the complete test plan in a private git repository before launching Claude, so a bare slash command cannot review the live workspace while a delayed fixture message remains queued. Preserve count thresholds, parsers and budgets. Add initial-context and installed-discovery tests, and retain startup/terminal diagnostics on failed evaluations.

* test: stabilize review fixtures and Claude eval startup

Preserve source boundaries in workflow judge inputs, isolate CEO mode plans, and wait for interactive trust input readiness. Keep startup failure evidence and retain existing models, budgets, and assertions.

Co-Authored-By: OpenAI Codex <noreply@openai.com>

* test: classify collapsed review modes and isolate seeded findings

Keep review questions out of the setup count when terminal cursor positioning removes spaces. State existing webhook safeguards so the five-finding control measures its seeded defects without accidental extra security and concurrency gaps. Preserve question bands and the paired control.

Co-Authored-By: OpenAI Codex <noreply@openai.com>

* test: isolate browser daemon state across free shards

Co-Authored-By: OpenAI Codex <noreply@openai.com>

* test: stabilize native review counting and interactive navigation

Co-Authored-By: OpenAI Codex <noreply@openai.com>

* chore: prepare v1.82.0.0 release

Co-Authored-By: OpenAI Codex <noreply@openai.com>

* fix: eliminate browser and process-cleanup test flakes

Pin every CI surface to Bun 1.4.0 to avoid extra-stdio finalizers closing
reused live sockets. Add an isolated GC/listener regression that fails on
Bun 1.3.13, and prevent coordinated rollback to an affected CI runtime.

Check renderer cleanup against the render's own staging directory so
concurrent renders cannot invalidate the assertion. Make the no-pgrep
process-tree walk tolerate disappearing /proc entries, and synchronize
its test fixture through child readiness and pipe EOF instead of sleeps.

Validation: 9,157 passed, 31 skipped, zero failures across 556 files with
retries disabled. Build, all-host generation freshness, and skill checks
passed. All three races have failing-before/passing-after regressions.

* fix: count completed native review questions in evals

* fix: drive review navigation from confirmed native choices

* fix: require complete section-loading eval reports

* test: isolate telemetry HTTP transport from local assertions

* fix: keep review input on the active native question

* test: let tunnel revocation daemon choose an available port

* test: allocate available ports for pairing and watchdog fixtures

* fix: stabilize planning eval navigation and phase reporting

* test: isolate installed runtime paths in planning evals

* test: stabilize review evidence and concurrent refresh fixtures

* fix: resolve design findings before editing the plan

* fix: honor and persist disabled outside plan reviews

* fix: preserve planning decisions and terminal evidence

Load installed host reviews at autoplan phase entry and wait for completed
reviewers and saved artifacts. Reuse approved remedies while preserving
individual finding decisions.

Drive interactive evals from the current terminal viewport, bind native
questions across scrolling, and require complete native report evidence.
Cover captured stale menus, permission lifecycles, setup classification,
and disabled-review tool availability with deterministic regressions.

Advance release metadata and the upgrade migration to the unclaimed
1.83.0.0 slot.

* fix: drive native review questions and preserve current plans

Use the native single-choice keyboard protocol and current terminal viewport,
with per-question navigation inside packets and completed-call coverage.
Keep permissions, multi-select menus, and Submit controls distinct.

Send Autoplan reviewers the amended implementation plan, keep its review record
separate, and supply retained application contracts in the chain fixture.
Clarify individual DevEx decisions and complete CEO fix options; use one active
plan destination for the section-loading report.

* fix: preserve complete plan-review decisions

* fix: recognize native plan dialogs and reviewer controls

* fix: preserve review decisions and phase completion

* fix: recognize completed reviews without losing findings

* fix: preserve review continuity and native eval completion

* test: fix native review completion and eval retry isolation

* test: handle native review menus and complete eval fixtures

* test: fix native review setup, completion, and isolation failures

* test: limit native skill discovery to runtime assets

* fix: bind Autoplan reviews to full ordered phase inputs

* test: fix planning eval routing, counting, and timeout handling

* chore: advance queued release to v1.84.0.0

* fix: preserve complete review inputs and planning decisions

* fix: reconcile review approvals and preserve phase obligations

* fix: preserve review obligations and unblock eval permissions

Carry recorded Autoplan requirements into blind phase inputs, require Eng
review approvals before exit, and exercise combined asynchronous flows in
CEO reviews. Correct native finding and handoff classification and unblock
repeated report edits using scoped request identities.

* fix: retain plan requirements and complete native review dialogs

* fix: complete native review prompts and retain plan references

* fix: preserve review inputs and classify native eval evidence

* fix: check competing completion orders in CEO reviews

* fix: recognize review decisions and require phase methodology

Require the current phase methodology before Autoplan snapshots. Correct
substantive decision, closed handoff, and cache-finding classification, and
honor the recommended implementation approach in native review dialogs.

Add captured-transcript regressions without changing review thresholds,
provider models, retries, or deadlines.

* test: bind native review decisions and close completed handoffs

* fix: complete review dialogs and verify methodology delivery

* fix: preserve review evidence and unblock native eval prompts

* fix: handle native review question completions

* fix: recognize native review narration and controls

* fix: count native review decisions and isolate eval fixtures

* test: verify seeded review coverage and current artifact permissions

* test: isolate model and brain-aware skill renders

* fix: repair native workflow evaluation and clarify review steps

* fix: stabilize workflow eval evidence and review guidance

* test: repair native workflow observation and fixture isolation

* fix: recognize completed workflow evidence and owned skill reads

* test: repair seeded workflow delivery and completion evidence

* test: recognize current review evidence across native forms

* test: handle native review variants and permission redraws

* fix: honor review preferences and recognize native eval evidence

* test: recognize completed review decisions and queued permissions

* test: match current review contracts and partial-line edits

* test: recognize completed workflow evidence and bounded human waits

* fix: preserve review entry gates and native eval interactions

* fix: recognize native workflow evidence and preserve review gates

* test: recognize current review evidence and preconfigure workflow fixtures

* test: recognize completed review findings and scoped artifact permissions

* fix: stabilize native workflow review and permission evidence

* fix: recognize current review evidence and scoped edit confirmations

Clarify Design and engineering review entry instructions and Design scoring.
Recognize required legacy coverage and public Autoplan completion recaps.
Bind the pending Edit confirmation to its exact file, ordered digest, and
one-request approval when a preceding command display remains visible.
Keep reviews within their existing size limits and preserve scope gates
when extracting workflow fixtures from either supported preamble header.

Keep failure outcomes, review thresholds, provider choices, and eval budgets.

* fix: recover review workflow progress and eval evidence

* fix: recognize valid review evidence and scope selection

* test: fix review evidence parsing and repeated artifact prompts

* test: recognize valid review decisions and pending native cards

* fix(plan-eng-review): keep final navigation consistent with approved tasks

* test: recognize valid review evidence and bind legacy diff requests

* fix: stabilize review eval evidence and harness repair guidance

* docs: update project documentation for v1.85.0.0

Co-Authored-By: OpenAI Codex <noreply@openai.com>

* test: fix Windows CI fixtures and credential scan

Rebase captured JSON values and filesystem evidence using the appropriate
path convention. Compile native fake CLIs on Windows and synchronize pipe
holder readiness, with cleanup retained when assertions fail.

Assemble synthetic credential fixtures at runtime so the added-line scan
keeps enforcing the same gate without flagging its own rejection controls.

Discover generated skills directly for the empty-find regression check,
avoiding a recursive scan through saved evaluation artifacts and dependencies.

* fix: preserve source renders on Windows

Compare canonical generator paths using native separators so an output
sidecar pointing at the source cannot overwrite its skill or metadata.
Keep the regression fixture isolated from the real checkout and expose
freshness diagnostics before asserting subprocess status.

Detach Windows drain-test pipe holders from the fake provider's automatic
child cleanup while preserving the enclosing runner job and its assertions.

* fix: clarify outside review fallback and CEO decisions

Render one applicable own-harness fallback path and retain native review,
disabled policy, and missing-coverage semantics. Align report field names
and mode labels, and make the existing per-cut scope approval explicit.

Regenerate skill outputs and keep the workflow judge's model, thresholds,
and retry policy unchanged.

* chore: move release to free version slot (v1.86.0.0)

PR #2852 now claims v1.85.0.0. Align the release metadata and
rename migration so upgrades from that version still receive it.

Co-Authored-By: OpenAI Codex <noreply@openai.com>

* fix: include engineering review prerequisites and restore branch context

* fix: recognize coverage diagrams and clarify design review instructions

* fix: preserve file identities and join Windows test processes

---------

Co-authored-by: OpenAI Codex <noreply@openai.com>
This commit is contained in:
Garry Tan
2026-09-14 14:32:45 -07:00
committed by GitHub
co-authored by OpenAI Codex
parent 71f6048e8a
commit 9f81911136
767 changed files with 127420 additions and 3917 deletions
Download Patch File Download Diff File Expand all files Collapse all files
claude/SKILL.md.tmpl
-347
View File
@@ -1,347 +0,0 @@
---
name: claude
preamble-tier: 3
version: 1.0.0
description: |
Claude Code CLI wrapper for non-Claude hosts - three modes. Review: independent
diff review via claude -p. Challenge: adversarial failure-mode review. Consult:
ask Claude about the repo with read-only file tools. Use when asked for "claude
review", "claude challenge", "ask claude", "second opinion from claude", or
"outside voice". (gstack)
triggers:
- claude review
- claude challenge
- ask claude
allowed-tools:
- Bash
- Read
- AskUserQuestion
---
{{PREAMBLE}}
{{BASE_BRANCH_DETECT}}
# /claude - Claude Outside Voice
You are running the `/claude` skill from a non-Claude host. This wraps `claude -p`
to get an independent Claude Code second opinion without allowing nested Claude to
modify files.
The generated external invocation name is `gstack-claude`.
Claude outside voices default to the frontier model via `{{CLAUDE_MODEL_FLAG}}`.
Use `GSTACK_CLAUDE_MODEL=<model>` to override it. If the user names a model in
the request, replace the default `--model` value with their model for every call,
including resumed sessions.
---
## Step 0: Resolve Claude CLI
```bash
CLAUDE_BIN=$(command -v claude 2>/dev/null || echo "")
[ -z "$CLAUDE_BIN" ] && echo "NOT_FOUND" || echo "FOUND: $CLAUDE_BIN"
```
If `NOT_FOUND`, stop and tell the user:
"Claude CLI not found. Install Claude Code, then re-run this skill."
Do not infer authentication state from credential files or environment variables.
Claude Code may use an OS keychain that is unavailable inside the host agent's
sandbox. On hosts that sandbox shell execution, run the actual `claude -p`
invocation outside that sandbox using the host's normal approval mechanism. Only
report an authentication blocker when that actual invocation returns an auth,
login, or unauthorized error.
Resolve the binary and invoke it in the same host execution context. Do not
resolve it inside a sandbox and then run a different `claude` from another PATH.
---
## Safety Boundary
Nested Claude must stay focused on the user's repository and must not run gstack
skills from inside this skill.
All `claude -p` calls MUST include:
- `--disable-slash-commands`
- Review/challenge: `--tools ""`
- Consult: `--allowedTools Read,Grep,Glob --disallowedTools Bash,Edit,Write`
Never pass `Bash`, `Edit`, or `Write` to nested Claude in this skill.
All prompts MUST be written to a temp file and fed through stdin. Never interpolate
user text directly into the shell command.
---
## Step 1: Detect Mode
Parse the user's input:
1. `/claude review` or `/claude review <instructions>` - **Review mode** (Step 2A)
2. `/claude challenge` or `/claude challenge <focus>` - **Challenge mode** (Step 2B)
3. `/claude` with no arguments, or `/claude <anything else>` - **Consult mode** (Step 2C)
If no mode is obvious and a diff exists, ask whether to review, challenge, or consult.
---
## Shared Helpers
Use these shell snippets in every mode.
Create temp files:
```bash
PROMPT_FILE=$(mktemp /tmp/gstack-claude-prompt-XXXXXX)
RESP_FILE=$(mktemp /tmp/gstack-claude-response-XXXXXX)
ERR_FILE=$(mktemp /tmp/gstack-claude-error-XXXXXX)
```
Cleanup at the end of every mode:
```bash
rm -f "$PROMPT_FILE" "$RESP_FILE" "$ERR_FILE"
```
Parse JSON output:
```bash
python3 - "$RESP_FILE" <<'PY'
import json, sys
path = sys.argv[1]
try:
obj = json.load(open(path))
except Exception as exc:
print(f"CLAUDE_JSON_PARSE_ERROR: {exc}")
sys.exit(0)
if obj.get("is_error"):
print("CLAUDE_ERROR: true")
result = obj.get("result") or obj.get("response") or ""
if result:
print(result)
usage = obj.get("usage") or {}
input_tokens = usage.get("input_tokens", 0) or 0
output_tokens = usage.get("output_tokens", 0) or 0
cache_read = usage.get("cache_read_input_tokens", 0) or 0
model = obj.get("model") or "unknown"
session_id = obj.get("session_id") or ""
print(f"\nTokens: input={input_tokens} output={output_tokens} cache_read={cache_read} | Model: {model}")
if session_id:
print(f"SESSION_ID:{session_id}")
PY
```
If stderr contains `auth`, `login`, or `unauthorized`, tell the user:
"Claude authentication failed. Run `claude` interactively to authenticate or export `ANTHROPIC_API_KEY`."
---
## Step 2A: Review Mode
Review the current branch diff with nested Claude in tool-less mode.
1. Fetch base and capture diff:
```bash
_REPO_ROOT=$(git rev-parse --show-toplevel) || { echo "ERROR: not in a git repo" >&2; exit 1; }
cd "$_REPO_ROOT"
DIFF_FILE=$(mktemp /tmp/gstack-claude-diff-XXXXXX)
git fetch origin <base> --quiet 2>/dev/null || true
git diff "origin/<base>" > "$DIFF_FILE" 2>/dev/null || git diff "<base>" > "$DIFF_FILE"
```
If the diff file is empty, stop and say:
"Nothing to review - no changes against the base branch."
2. Write the prompt file:
```bash
cat > "$PROMPT_FILE" <<'EOF'
You are a brutally honest Claude Code reviewer. Review this git diff for bugs,
production failure modes, security issues, missing tests, and maintainability
problems. Be direct. No compliments. Reference files and changed code where possible.
Additional user instructions, if any:
<custom review instructions>
DIFF:
EOF
cat "$DIFF_FILE" >> "$PROMPT_FILE"
```
3. Run Claude:
```bash
CLAUDE_BIN=$(command -v claude 2>/dev/null) || { echo "Claude CLI not found" >&2; exit 1; }
cat "$PROMPT_FILE" | "$CLAUDE_BIN" -p {{CLAUDE_MODEL_FLAG}} --output-format json --disable-slash-commands --tools "" > "$RESP_FILE" 2>"$ERR_FILE"
```
4. Present the parsed output:
```
CLAUDE SAYS (code review):
============================================================
<parsed result from RESP_FILE>
============================================================
```
5. Cleanup:
```bash
rm -f "$DIFF_FILE" "$PROMPT_FILE" "$RESP_FILE" "$ERR_FILE"
```
---
## Step 2B: Challenge Mode
Run an adversarial failure-mode review with nested Claude in tool-less mode.
1. Capture the diff using the same diff commands from Review mode.
2. Write the prompt:
```bash
cat > "$PROMPT_FILE" <<'EOF'
You are an adversarial Claude Code reviewer. Try to break this change before users do.
Find edge cases, race conditions, security holes, resource leaks, silent data
corruption, bad error handling, and operational failure modes. Be thorough. No
compliments. If the user provided a focus area, prioritize it.
Focus area, if any:
<focus>
DIFF:
EOF
cat "$DIFF_FILE" >> "$PROMPT_FILE"
```
3. Run Claude:
```bash
CLAUDE_BIN=$(command -v claude 2>/dev/null) || { echo "Claude CLI not found" >&2; exit 1; }
cat "$PROMPT_FILE" | "$CLAUDE_BIN" -p {{CLAUDE_MODEL_FLAG}} --output-format json --disable-slash-commands --tools "" > "$RESP_FILE" 2>"$ERR_FILE"
```
4. Present the parsed output:
```
CLAUDE SAYS (adversarial challenge):
============================================================
<parsed result from RESP_FILE>
============================================================
```
5. Cleanup:
```bash
rm -f "$DIFF_FILE" "$PROMPT_FILE" "$RESP_FILE" "$ERR_FILE"
```
---
## Step 2C: Consult Mode
Ask Claude about the repository. Consult mode may inspect files, but only with
read-only tools.
1. Check for an existing Claude session:
```bash
cat .context/claude-session-id 2>/dev/null || echo "NO_SESSION"
```
If a session exists, ask the user whether to continue it or start fresh.
2. Write the prompt:
```bash
cat > "$PROMPT_FILE" <<'EOF'
You are Claude Code acting as an independent outside voice for this repository.
Answer the user's question directly. You may inspect repository files with Read,
Grep, and Glob only. Do not use Bash. Do not edit or write files. Do not invoke
slash commands or gstack skills.
USER QUESTION:
<user prompt>
EOF
```
3. Run Claude.
For a new session:
```bash
CLAUDE_BIN=$(command -v claude 2>/dev/null) || { echo "Claude CLI not found" >&2; exit 1; }
cat "$PROMPT_FILE" | "$CLAUDE_BIN" -p {{CLAUDE_MODEL_FLAG}} --output-format json --disable-slash-commands --allowedTools Read,Grep,Glob --disallowedTools Bash,Edit,Write > "$RESP_FILE" 2>"$ERR_FILE"
```
For a resumed session:
```bash
CLAUDE_BIN=$(command -v claude 2>/dev/null) || { echo "Claude CLI not found" >&2; exit 1; }
cat "$PROMPT_FILE" | "$CLAUDE_BIN" -p --resume "<session-id>" {{CLAUDE_MODEL_FLAG}} --output-format json --disable-slash-commands --allowedTools Read,Grep,Glob --disallowedTools Bash,Edit,Write > "$RESP_FILE" 2>"$ERR_FILE"
```
4. Parse and save the session id:
```bash
SESSION_ID=$(python3 - "$RESP_FILE" <<'PY'
import json, sys
try:
obj = json.load(open(sys.argv[1]))
print(obj.get("session_id") or "")
except Exception:
print("")
PY
)
if [ -n "$SESSION_ID" ]; then
mkdir -p .context
printf "%s\n" "$SESSION_ID" > .context/claude-session-id
fi
```
5. Present the parsed output:
```
CLAUDE SAYS (consult):
============================================================
<parsed result from RESP_FILE>
============================================================
Session saved - run /claude again to continue this conversation.
```
6. Cleanup:
```bash
rm -f "$PROMPT_FILE" "$RESP_FILE" "$ERR_FILE"
```
---
## Error Handling
- **Binary not found:** Stop with install instructions.
- **Auth failure from the actual host invocation:** Stop with login/API key instructions.
- **Auth failure from stderr:** Surface the stderr line and ask the user to re-authenticate.
- **JSON parse failure:** Show raw stdout from `$RESP_FILE` and stderr from `$ERR_FILE`.
- **Empty response:** Tell the user "Claude returned no response. Check stderr for errors."
- **Resume failure:** Delete `.context/claude-session-id` and retry with a fresh session.
---
## Important Rules
- Nested Claude is read-only in consult mode and tool-less in review/challenge.
- Always include `--disable-slash-commands`.
- Never pass nested Claude `Bash`, `Edit`, or `Write`.
- Never interpolate user text into a shell command.
- Present Claude's response faithfully, then add any host-agent synthesis after it.