Files
gstack/test/state-root-ratchet.test.ts
ac20ef1e13 v1.91.16.0 fix: safe installs and upgrades, skills start everywhere, truthful completion (parallel wave) (#3014)
* fix(make-pdf,design): real TOC page numbers and anchors; design models move to gpt-5.5 + gpt-image-2 (#2903, #2807)

* ci: pin every action to a SHA, Scorecard, checksummed Bun, scoped Codex CLI in the CI image, macOS named-regressions job (#1948, #1706, #1997)

* fix(cso,browse): Windows/macOS lease identity, retention skips bad runs, truthful 'not assessed' reports, env.kv false positive, Windows browse spawn fixes (#2894, #2912, #2048, #2637, #2805)

* fix(outside-voice): one Codex model precedence for probe and dispatch, bounded timeouts with KILL escalation, probe cache identity, guarded mktemp, skills hidden in nested Codex (#2776, #2914, #2787, #2881, #2847)

* feat(setup): install registry, ./setup --status, staged activation, per-scope migrations, host tiers and conformance kit, OpenCode/Hermes fixes, Copilot CLI (experimental), disabled_skills, slimmer CLAUDE.md (#1925, #2879, #2347, #2338, #2626, #2629, #2825, #2015, #393, #1206, #2096)

Co-authored-by: Andrey Esipov <61357035+andrey-esipov@users.noreply.github.com>
Co-authored-by: Matthew Williams <5863886+ridermw@users.noreply.github.com>
Co-authored-by: Rocky Wu <92764750+lolisaigao1234@users.noreply.github.com>

* fix(preamble): skills start in worktree-isolated Claude Code sessions — one literal skill-start line, context-recovery helper, no eval in skill blocks, $N token tripwire, honest plan-mode precedence (#2763, #2896, #2851)

* fix(setup): per-install renders for renamed, vendored and custom-config installs, space-safe alias roots, per-host /cso launcher path (#1882, #2906)

* test: free tests never write the real HOME (home-write tripwire), no orphaned browse daemons, symlinked-TMPDIR (macOS) fixes, gbrain-on-PATH isolation (#2895, #2816, #2829, #2841, #2982, #2911, #2909, #2908, #2910)

* chore: integrate wave follow-ups — regenerate all hosts, README model docs, TODOS, touchfiles, duration seeds, Bun 1.3 telemetry test fix, replay fixtures pinned to their recorded bytes

* v1.91.14.0: release metadata

* test(browse): group-kill cleanup falls back to the leader when macOS returns EPERM for a live group

* ci: Codex login steps run under bash (the CI container's default sh rejects pipefail)

* fix: full-suite repairs — private state roots for 7 real-HOME writers, truthful home-guard attribution with --attribute-home, gstack-paths shim in ship-hook fixtures, Copilot in team-init probes, test-of-test detector, cookie review fixture pinned to the approved excerpt

* fix(windows): LF for Dockerfiles, normalized home-guard exclusions, path.relative fixture containment, ship push test spawn timeouts

* test(design-detect): widen the whole-scan budget test's timing margins so a loaded macOS runner can't time out the single-batch call

* fix(evals): qa protocol names defined for the judge, async PTY judge so overlapping sessions keep their cleanup deadline, bws setup cases use the standard JUDGE_MS tier

* test(pty): the non-blocking judge case is POSIX-only (its fake judge is a shebang script, like the PTY harness it covers)

* test(evals): ship-coverage-audit names absolute skill paths, uses Read, and skips the Step 7 artifact nothing consumes (removes the mechanism behind 3 of 4 historical timeouts)

* test: deflake terminal-agent concurrent startup (write the placeholder record before spawning, like the product) and qa-evidence resumed pipe (barrier on a full high-water mark instead of a sleep)

* fix(design): daemon log stream handles open errors instead of crashing the daemon; shutdown test awaits the stubbed exit

* test(evals): skillify checks the product's resolved state root (no GSTACK_HOME bypass), negative cases check every root; tpa shims live outside the agent's cwd and answer consistently

* test(evals): coverage-audit detector splits parallel columns at │ rails and [+] groups; coverage prompts name absolute skill paths; TODOs from the gate eval census

* fix(plan-eng-review): scope gate always sends the menu (no headless 'Scope pending' exit); PTY harness ends idle turn-ends and API-error panels promptly with a precise diagnostic

* test(evals): one position-bound coverage-read rule credits read-only && chains (labels, context ranges, git displays in either order) and still rejects forged output

* fix(ship): documentation section has one saved source inserted by cat (never retyped); spawned-marker evidence is the child output; refreshed codex/factory ship goldens

* test(docsync): the byte-exact insert positive is POSIX-only (the bash-command observer refuses backslash paths by design)

* test(evals): plan-design-with-ui focus-title rule accepts the comma-less 'dimensions or focus' wording (replayed on the captured CI run, negative controls kept)

* fix(qa-callers): exploratory preflight says the scope/method Reads come after exploratory.md; design-review scan check credits an executed scan proven by one engine run over the changed files

---------

Co-authored-by: Andrey Esipov <61357035+andrey-esipov@users.noreply.github.com>
Co-authored-by: Matthew Williams <5863886+ridermw@users.noreply.github.com>
Co-authored-by: Rocky Wu <92764750+lolisaigao1234@users.noreply.github.com>
2026-10-03 07:54:09 -07:00

167 lines
8.7 KiB
TypeScript

/**
* Ratchet (a): one state-root rule. Outside the two owners (lib/state-root.ts,
* bin/gstack-state-root.sh) and the committed allowlist, no tracked file may
* hand-roll the chain, and every gstack-paths eval must fail stop.
*/
import { describe, test, expect } from 'bun:test';
import { spawnSync } from 'child_process';
import * as fs from 'fs';
import * as path from 'path';
import { resolveStateRoot } from '../lib/state-root';
const ROOT = path.resolve(import.meta.dir, '..');
const ALLOWLIST_PATH = 'test/state-root-ratchet.allowlist.json';
type AllowEntry = { path: string; match: string; reason?: string };
type Violation = { file: string; line: number; text: string; rule: string };
const CHAIN = /\\?\$\{GSTACK_(?:HOME|STATE_ROOT|STATE_DIR):-/;
const HOMEDIR_GSTACK = /homedir\(\)\s*,\s*['"`]\.gstack['"`]/g;
const PROSE_ROOT = /(~|\\?\$HOME|\\?\$\{HOME\})\/\.gstack(?![\w-])/;
const PATHS_EVAL = /eval "\$\(.*gstack-paths"?\)"/;
const PATHS_GET = /\$\(\S*gstack-paths"? --get GSTACK_STATE_ROOT\)/;
const GUARD = ': "${GSTACK_STATE_ROOT:?gstack-paths failed; reinstall with ./setup or /gstack-upgrade}"';
const GUARD_TS = ': "\\${GSTACK_STATE_ROOT:?gstack-paths failed; reinstall with ./setup or /gstack-upgrade}"';
const isProseSource = (f: string) => f.endsWith('.tmpl') || f.startsWith('scripts/resolvers/');
const isExecutable = (f: string, text: string) =>
/^#!.*\b(bash|sh)\b/.test(text) || f.endsWith('.sh') || f === 'setup';
/** Lines inside ```bash / ```sh fences (template-literal fences \`\`\` included), comments skipped. */
function bashBlockLines(text: string): Array<[number, string]> {
const out: Array<[number, string]> = [];
let inBash = false;
text.split('\n').forEach((line, i) => {
const fence = line.trim().match(/^(?:\\`|`){3}(\w*)/);
if (fence) { inBash = inBash ? false : /^(bash|sh)$/.test(fence[1]); return; }
if (inBash && !line.trim().startsWith('#')) out.push([i + 1, line]);
});
return out;
}
export function scan(files: Map<string, string>): Violation[] {
const v: Violation[] = [];
for (const [file, text] of files) {
const lines = text.split('\n');
const code = (line: string) => !/^\s*(#|\/\/|\*)/.test(line) || isProseSource(file);
lines.forEach((line, i) => {
if (code(line) && CHAIN.test(line)) v.push({ file, line: i + 1, text: line.trim(), rule: 'inline state-root chain' });
});
if (/\.(ts|js|mjs)$/.test(file) || !file.includes('.')) {
for (const m of text.matchAll(HOMEDIR_GSTACK)) {
const line = text.slice(0, m.index).split('\n').length;
v.push({ file, line, text: lines[line - 1].trim(), rule: 'homedir() joined with .gstack' });
}
}
if (isProseSource(file)) {
for (const [n, line] of bashBlockLines(text)) {
if (PROSE_ROOT.test(line)) v.push({ file, line: n, text: line.trim(), rule: 'executable ~/.gstack in a bash block' });
}
lines.forEach((line, i) => {
if (/export GSTACK_STATE_ROOT\b/.test(line)) v.push({ file, line: i + 1, text: line.trim(), rule: 'export GSTACK_STATE_ROOT in prose' });
});
}
if (isProseSource(file) || isExecutable(file, text)) {
const guarded = (line: string) => line.includes(GUARD) || line.includes(GUARD_TS);
lines.forEach((line, i) => {
if (!code(line)) return;
// Skill blocks never eval (#2763): worktree-isolated Claude Code
// sessions refuse it, guarded or not. Executables may.
if (isProseSource(file) && PATHS_EVAL.test(line)) {
v.push({ file, line: i + 1, text: line.trim(), rule: 'gstack-paths eval in a skill block' });
} else if ((PATHS_EVAL.test(line) || PATHS_GET.test(line)) && !guarded(line)) {
v.push({ file, line: i + 1, text: line.trim(), rule: 'unguarded gstack-paths eval' });
}
});
}
}
return v;
}
function allowed(v: Violation, allow: AllowEntry[]): boolean {
return allow.some((a) => (a.path.endsWith('/') ? v.file.startsWith(a.path) : v.file === a.path)
&& (a.match === '*' || v.text === a.match));
}
function report(violations: Violation[]): string {
if (violations.length === 0) return '';
return [
...violations.map((x) => `${x.file}:${x.line} ${x.text} (${x.rule})`),
'Rule: gstack state lives under one root resolved by one chain; a hand-rolled chain drifts (plugin installs and GSTACK_HOME users silently split their state), and an unguarded eval writes under / when the resolver is missing.',
'Fix: in skill prose use GSTACK_STATE_ROOT=$(~/.claude/skills/gstack/bin/gstack-paths --get GSTACK_STATE_ROOT); ' + GUARD
+ ' (worktree-isolated Claude Code sessions refuse eval); in executables use eval "$(~/.claude/skills/gstack/bin/gstack-paths)"; ' + GUARD
+ ' (hooks and bins may source bin/gstack-state-root.sh instead); in TS use resolveStateRoot() from lib/state-root.ts.',
`Allowlist: ${ALLOWLIST_PATH}, keyed on file path plus matched line text, each entry with a reason. Legitimate only for illustrative prose, an install location that is not state, or a named partial-upgrade fallback.`,
].join('\n');
}
function trackedFiles(): Map<string, string> {
const r = spawnSync('git', ['ls-files', '-z'], { cwd: ROOT, encoding: 'utf-8', timeout: 30_000 });
const files = new Map<string, string>();
for (const f of r.stdout.split('\0').filter(Boolean)) {
if (/\.(png|jpg|jpeg|gif|webp|pdf|ico|woff2?|ttf|lock|onnx|bin|zip)$/i.test(f)) continue;
const full = path.join(ROOT, f);
try {
const stat = fs.lstatSync(full);
if (!stat.isFile() || stat.size > 2_000_000) continue;
files.set(f, fs.readFileSync(full, 'utf-8'));
} catch { /* deleted in the working tree */ }
}
return files;
}
const allow: AllowEntry[] = JSON.parse(fs.readFileSync(path.join(ROOT, ALLOWLIST_PATH), 'utf-8'));
describe('ratchet (a): one state-root rule', () => {
test('every allowlist entry carries a reason', () => {
const missing = allow.filter((a) => !a.reason || !a.reason.trim());
expect(missing.map((a) => `${a.path} :: ${a.match}`)).toEqual([]);
});
test('no tracked file outside the owners and the allowlist hand-rolls the state root', () => {
expect(report(scan(trackedFiles()).filter((x) => !allowed(x, allow)))).toBe('');
});
test('planted violations fail with file:line, the Fix: text and the allowlist path', () => {
const planted = new Map<string, string>([
['bin/gstack-planted', '#!/usr/bin/env bash\n# a comment "${GSTACK_HOME:-$HOME/.gstack}" is prose\nX="${GSTACK_HOME:-$HOME/.gstack}"\neval "$("$(dirname "$0")/gstack-paths")"\n'],
['lib/planted.ts', "import { homedir } from 'os';\nconst r = join(\n homedir(), '.gstack');\n"],
['planted/SKILL.md.tmpl', 'Prose ~/.gstack is fine.\n```bash\n# comment ~/.gstack is fine\nmkdir -p ~/.gstack/analytics\nexport GSTACK_STATE_ROOT\neval "$(~/.claude/skills/gstack/bin/gstack-paths)"\neval "$(~/.claude/skills/gstack/bin/gstack-paths)"; ' + GUARD + '\nGSTACK_STATE_ROOT=$(~/.claude/skills/gstack/bin/gstack-paths --get GSTACK_STATE_ROOT)\n```\n'],
]);
const found = scan(planted);
expect(found.map((x) => `${x.file}:${x.line}:${x.rule}`)).toEqual([
'bin/gstack-planted:3:inline state-root chain',
'bin/gstack-planted:4:unguarded gstack-paths eval',
'lib/planted.ts:3:homedir() joined with .gstack',
'planted/SKILL.md.tmpl:4:executable ~/.gstack in a bash block',
'planted/SKILL.md.tmpl:5:export GSTACK_STATE_ROOT in prose',
'planted/SKILL.md.tmpl:6:gstack-paths eval in a skill block',
'planted/SKILL.md.tmpl:7:gstack-paths eval in a skill block',
'planted/SKILL.md.tmpl:8:unguarded gstack-paths eval',
]);
const message = report(found);
expect(message).toContain('planted/SKILL.md.tmpl:4 mkdir -p ~/.gstack/analytics');
expect(message).toContain('Fix: ');
expect(message).toContain('resolveStateRoot()');
expect(message).toContain(ALLOWLIST_PATH);
});
test('allowlist entries are keyed on text, not line numbers: inserting a line above still passes', () => {
const file = 'careful/bin/check-careful.sh';
const text = fs.readFileSync(path.join(ROOT, file), 'utf-8');
const shifted = new Map([[file, `# inserted line\n${text}`]]);
const found = scan(shifted);
expect(found.length).toBeGreaterThan(0);
expect(found.filter((x) => !allowed(x, allow))).toEqual([]);
});
test('guarded evals and the owners pass, and the TS owner resolves the documented default', () => {
const ok = new Map([
['x/SKILL.md.tmpl', '```bash\nGSTACK_STATE_ROOT=$(~/.claude/skills/gstack/bin/gstack-paths --get GSTACK_STATE_ROOT); ' + GUARD + '\nls "$GSTACK_STATE_ROOT"/projects\n```\n'],
['bin/gstack-x', '#!/usr/bin/env bash\neval "$("$(dirname "$0")/gstack-paths")"; ' + GUARD + '\n'],
]);
expect(scan(ok)).toEqual([]);
expect(resolveStateRoot({ HOME: '/h' }, 'linux')).toBe('/h/.gstack');
});
});