fix: round 3 of the release review of 6.3.0

Round 3 (v6.2.0..96cb977, BLOCK: 1 BLOCKER, 8 MAJOR, 2 MINOR), each fixed
behind a test that failed first.

- Every allowlisted runner, not only git, takes plain words and a short list
  of options that only read: an option that writes a file or loads code
  (--test-reporter-destination, --outputFile, --junitxml, --basetemp,
  --import, --config, -p, -o, -exec, a reporter path) is on no list; after
  `npm test --` only paths follow; `make test` takes no VAR=value. A built-in
  node test reporter (tap, spec, dot, junit, lcov) stays allowed.
- /trekexecute: a restore the guard refuses (destructive git, in the mode
  that shows no prompts) or the operator declines is never recorded as
  reverted: simplify = "revert-refused" and the run stops before review; a
  refused step revert is reported with its paths.
- stepTestsOnly: a fixture document under tests/ is a test file, "e.g." is
  not a path, ./ is dropped; the plan-file sweep reads Files: with the same
  reader (fieldPaths); the validator hands its parsed fields over.
- plan-validator --steps-from asks for Verify: on a 1.8 plan too.
- skill frontmatter: a value starting with # is a hazard; a quote after an
  escaped backslash closes the value. Routing reads kind tags with hasOwn.
- release-check reads scope_sha_end as written (an all-digit sha keeps its
  leading zero), and Voyage's own artifacts under .claude/projects/ and
  .claude/voyage/ committed after the review do not count as unreviewed.

Suite 2170 (2168/0/2).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

commands/trekexecute.md

@@ -1863,6 +1863,12 @@ directly (Sub-step B).
   ```
   The new files stay on disk, untracked, and `step-{N}-upward.md` names them —
   nothing the step made is deleted, and none of it is left registered.
+  When the guard refuses the restore (it is destructive git: in the permission
+  mode that shows no prompts the guard refuses it, exit 2, "no one to ask"), or
+  the operator declines it, nothing was restored: write in `step-{N}-upward.md`
+  and the final report that the step's changes are still in the working tree,
+  with the paths, and that the operator restores them by hand. Never call the
+  step reverted.
   Record failure. **Do NOT proceed to next step.** Jump to Phase 7.5 (Phase 7
   runs only when every step passed).
 
@@ -2177,6 +2183,14 @@ never typed by the operator, in this order (Cherny's):
      `refactor: simplify pass (trekexecute)`; `simplify = "applied"`. Exit 1 or
      2 → restore every path it changed as above, `simplify = "reverted"`, and
      say why in the report: the cleanup broke the verification.
+   - **A restore did not run** (the guard refused it — `git checkout --` is
+     destructive git, refused in the permission mode that shows no prompts —
+     or the operator declined it): the paths are still changed.
+     `simplify = "revert-refused"`, never `"reverted"`. When the cleanup broke
+     the verification, stop here: the run is `partial`, no review runs over a
+     tree that failed it, and the report lists the paths for the operator to
+     restore by hand. Paths outside the run's files that were not restored are
+     listed too.
    - **It could not start** (the Skill tool refused, the skill is not there):
      `simplify = "not-run"`, with the reason. Not a stop: `/simplify` is a
      cleanup pass, not a gate. Continue.
@@ -2855,7 +2869,7 @@ To continue (multi-session waves): re-run the same /trekexecute command (not --r
     "steps_check_not_measured": 0,
     "arch_rules": "{pass | violation | not-measured | n/a}",
     "scope_sweep": "{pass | deviation | not-measured | n/a}",
-    "simplify": "{applied | clean | reverted | not-run | n/a}",
+    "simplify": "{applied | clean | reverted | revert-refused | not-run | n/a}",
     "review_verdict": "{ALLOW | WARN | BLOCK | none}",
     "review_gate_exit": "{0 | 1 | 2}",
     "main_merge": "{merged | nothing-to-merge | closed | declined | n/a}"

docs/HANDOVER-CONTRACTS.md

@@ -375,7 +375,7 @@ contract CLI's exit said in words; `not-measured` / `not-run` are never green.
 | `plan_approval` | run | `approved \| not-checked` | K2 `approval-log.mjs verify --kind plan` (Phase 1); `not-checked` = a session spec |
 | `arch_rules` | run | `pass \| violation \| not-measured` | K5 `lib/verification/arch-rules.mjs` exit 0 / 1 / 2 (Phase 7) |
 | `scope_sweep` | run | `{status: pass \| deviation \| not-measured, outside: [paths]}` | K6 `lib/verification/plan-scope-sweep.mjs` exit 0 / 1 / 2 (Phase 7) |
-| `simplify` | run | `applied \| clean \| reverted \| not-run` | `/simplify` in Phase 7.2 |
+| `simplify` | run | `applied \| clean \| reverted \| revert-refused \| not-run` | `/simplify` in Phase 7.2; `revert-refused` = the restore was refused or declined |
 | `review` | run | `{round ≥ 1, path, gate_exit: 0 \| 1 \| 2, verdict}` | K7 `lib/review/verdict-gate.mjs` on `review.md` (Phase 7.2; Phase 8 locks on it) |
 
 `plan_verification.failed_criteria[]` carries `exit_code` (integer, or `null`

docs/release-runbook.md

@@ -81,8 +81,9 @@ do. `scripts/release-check.mjs` refuses a release whose section has no
    It checks, in order: the working tree is clean (otherwise the suite would
    not measure HEAD), `npm test` exits 0 at HEAD, the review verdict is not
    BLOCK, the review covers HEAD (its `scope_sha_end` is HEAD or an ancestor,
-   and the commits after it touch only the release bookkeeping above), and the
+   and the commits after it touch only the release bookkeeping above and
-   door type is set. 0 = ready, 1 = not ready (each failure listed), 2 = could
+   Voyage's own artifacts under `.claude/projects/` and `.claude/voyage/`, such
+   as the review itself and the approval log), and the door type is set. 0 = ready, 1 = not ready (each failure listed), 2 = could
    not run — never green.
 6. **The operator's yes, logged.** The operator reads the release section and
    says yes (an answer like "sounds good" is not a yes). The check prints the

lib/parsers/plan-schema.mjs

@@ -293,7 +293,15 @@ export function onFailureAction(text) {
  * so a bare `Makefile` is still a path. Release review of 6.3.0, rounds 1-2.
  */
 const PATH_WORD = /^(?:[\w@.-]+\/)+[\w@.-]+$|^[\w@-][\w@.-]*\.[A-Za-z0-9]+$/;
-function fieldPaths(text) {
+// Abbreviations that PATH_WORD would read as a name with an extension.
+const NOT_A_PATH = /^(?:e\.g|i\.e|n\.b|cf|etc|vs)$/i;
+const dotSlash = (p) => p.replace(/^\.\//, '');
+/**
+ * The one reader of a step's `Files:` field: this file's tests-only check and
+ * the plan-file sweep (lib/verification/plan-scope-sweep.mjs) both use it, so
+ * the two never judge different lists for one step. A leading `./` is dropped.
+ */
+export function fieldPaths(text) {
   const spans = [];
   const masked = String(text ?? '').replace(/`([^`\n]+)`/g, (_, p) => `\u0000${spans.push(p.trim()) - 1}\u0000`);
   const out = [];
@@ -304,17 +312,20 @@ function fieldPaths(text) {
     for (const word of entry.split(/\s+/)) {
       const span = /^\u0000(\d+)\u0000$/.exec(word);
       if (span) found.push(spans[Number(span[1])]);
-      else if (PATH_WORD.test(word.replace(/[.;:]+$/, ''))) found.push(word.replace(/[.;:]+$/, ''));
+      else {
+        const bare = word.replace(/[.;:]+$/, '');
+        if (PATH_WORD.test(bare) && !NOT_A_PATH.test(bare)) found.push(bare);
+      }
     }
     out.push(...(found.length > 0 ? found : [entry.split(/\s+/)[0]]));
   }
-  return out.filter((p) => p && !p.includes('\u0000'));
+  return out.filter((p) => p && !p.includes('\u0000')).map(dotSlash);
 }
 
-// A document is never the test here: `docs/specs/api.md` sits under a `specs/`
+// A document or a data file outside tests/ is never the test here:
-// folder, which lib/util/test-path.mjs reads as tests for the floor guard.
+// `docs/specs/api.md` and `api/specs/openapi.yaml` sit under a `specs/` folder,
-// Nor is a data file under a `spec/` or `specs/` folder (`api/specs/openapi.yaml`
+// which lib/util/test-path.mjs reads as tests for the floor guard. Under tests/
-// is an API spec); under tests/ it is a fixture, and a fixture is a test file.
+// either one is a fixture, and a fixture is a test file.
 const DOC_RE = /\.(md|markdown|mdx|txt|rst|adoc)$/i;
 const DATA_RE = /\.(ya?ml|json|toml|xml|csv)$/i;
 const UNDER_TESTS_RE = /(^|\/)(tests?|__tests__)\//;
@@ -333,13 +344,17 @@ const UNDER_TESTS_RE = /(^|\/)(tests?|__tests__)\//;
  * @returns {{testsOnly: boolean, files: string[]}}
  */
 export function stepTestsOnly(body) {
-  const f = parseStepFields(body);
+  return stepTestsOnlyFromFields(parseStepFields(body));
+}
+
+/** `stepTestsOnly` for a caller that already holds the step's parsed fields. */
+export function stepTestsOnlyFromFields(f) {
   const files = f.has('Files') ? fieldPaths(f.get('Files').text) : [];
   const testFirst = f.has('Test first')
-    ? [...f.get('Test first').text.matchAll(/\bFile:\**[ \t]*`([^`\n]+)`/gi)].map((m) => m[1].trim())
+    ? [...f.get('Test first').text.matchAll(/\bFile:\**[ \t]*`([^`\n]+)`/gi)].map((m) => dotSlash(m[1].trim()))
     : [];
   const isTest = (p) => testFirst.includes(p) ||
-    (isTestPath(p) && !DOC_RE.test(p) && !(DATA_RE.test(p) && !UNDER_TESTS_RE.test(p)));
+    (isTestPath(p) && (UNDER_TESTS_RE.test(p) || !(DOC_RE.test(p) || DATA_RE.test(p))));
   return { testsOnly: files.length > 0 && files.every(isTest), files };
 }
 

lib/skills/routing.mjs

@@ -190,7 +190,7 @@ export function loadCases(root) {
     const text = readFileSync(f, 'utf8');
     const tagsField = readFrontmatterFields(text).fields.tags;
     const tags = tagsField && Array.isArray(tagsField.value) ? tagsField.value.map(String) : [];
-    const kindTag = tags.find((t) => t in KIND_TAGS);
+    const kindTag = tags.find((t) => Object.hasOwn(KIND_TAGS, t));
     if (!kindTag) continue;
     const val = (k) => { const t = tags.find((x) => x.startsWith(`${k}=`)); return t ? t.slice(k.length + 1) : null; };
     const body = text.replace(/^---\n[\s\S]*?\n---\n?/, '').trim();

lib/skills/skill-frontmatter.mjs

@@ -31,6 +31,7 @@ export function splitFile(text) {
 export function plainScalarHazards(value) {
   const out = [];
   if (/:\s/.test(value) || /:$/.test(value)) out.push('unquoted ": " (YAML reads a mapping and the line fails)');
+  if (/^#/.test(value)) out.push('"#" at the start is a YAML comment: the value is empty');
   if (/\s#/.test(value)) out.push('" #" starts a YAML comment: the rest of the value is cut');
   if (/^[`@%*&!|>,?{]/.test(value)) out.push(`a value may not start with "${value[0]}" unquoted`);
   if (/^\[/.test(value) && !/\]$/.test(value)) out.push('an unclosed flow list');
@@ -40,7 +41,9 @@ export function plainScalarHazards(value) {
 
 function unquote(v) {
   if (v.startsWith('"')) {
-    if (v.length < 2 || !v.endsWith('"') || v.endsWith('\\"')) return { error: 'unclosed double quote' };
+    // The last quote closes the value unless an odd run of backslashes escapes it.
+    const escaped = (v.slice(1, -1).match(/\\*$/)[0].length % 2) === 1;
+    if (v.length < 2 || !v.endsWith('"') || escaped) return { error: 'unclosed double quote' };
     try { return { value: JSON.parse(v) }; } catch { return { error: 'bad escape in a double-quoted value' }; }
   }
   if (v.startsWith("'")) {

lib/validators/plan-validator.mjs

@@ -40,7 +40,7 @@ import {
   sliceSteps, validatePlanHeadings, extractPlanVersion, extractPlanCeremony,
   parseStepFields, findSection, splitSections, findModuleCycle, parseMermaid, fencedBlocks,
   versionAtLeast, PLAN_VERSION_CURRENT, PLAN_VERSIONS_SUPPORTED, CEREMONY_VALUES,
-  ON_FAILURE_ACTIONS, onFailureAction, stepTestsOnly,
+  ON_FAILURE_ACTIONS, onFailureAction, stepTestsOnlyFromFields,
 } from '../parsers/plan-schema.mjs';
 import { validateAllManifests } from '../parsers/manifest-yaml.mjs';
 import { parseDocument } from '../util/frontmatter.mjs';
@@ -165,7 +165,7 @@ function checkRecovery(sections, opts) {
         `Step ${s.n}: \`On failure:\` does not open with a recovery word the executor reads`,
         `Open with one of ${ON_FAILURE_ACTIONS.join(' | ')}, then the specifics (\`retry once, then escalate\`).`));
     }
-    if (stepTestsOnly(s.body).testsOnly) {
+    if (stepTestsOnlyFromFields(f).testsOnly) {
       // The run-the-thing script alone is such a step too: `End-to-end:` rides on
       // the last step that builds behaviour (release review of 6.3.0, 2026-10-10).
       const e2e = f.has('End-to-end')
@@ -298,11 +298,13 @@ export function validatePlanContent(rawText, opts = {}) {
   // Phase 7.2) must be runnable steps whatever the plan's version: below 1.8
   // the step rules never ran, and a review step whose Verify is "re-run
   // /trekreview" validated and was executed (release review of 6.3.0,
-  // 2026-10-10). From 1.8 the rules below already cover every step.
+  // 2026-10-10). From 1.8 the rules below cover every step except one: none of
+  // them asks for a `Verify:`, so an appended step needs one on any version
+  // (round 3).
   const stepsFrom = Number.isInteger(opts.stepsFrom) && opts.stepsFrom > 0 ? opts.stepsFrom : null;
-  if (stepsFrom !== null && !versionAtLeast(planVersion, '1.8')) {
+  if (stepsFrom !== null) {
     const appended = sections.filter((s) => Number(s.n) >= stepsFrom);
-    const stepIssues = [
+    const stepIssues = versionAtLeast(planVersion, '1.8') ? [] : [
       ...checkSteps(appended).filter((i) => i.code !== 'PLAN_NO_END_TO_END'),
       ...checkRunnable(text, appended, { cwd: opts.cwd, stepsOnly: true }),
     ];

lib/validators/progress-validator.mjs

@@ -40,7 +40,7 @@ export const PROGRESS_REGRESSION_VALUES = ['passed', 'failed', 'unrunnable', 'no
 export const PROGRESS_PLAN_APPROVAL_VALUES = ['approved', 'not-checked']; // K2; not-checked = a session spec
 export const PROGRESS_ARCH_RULES_VALUES = ['pass', 'violation', 'not-measured']; // K5 exit 0 / 1 / 2
 export const PROGRESS_SCOPE_SWEEP_VALUES = ['pass', 'deviation', 'not-measured']; // K6 exit 0 / 1 / 2
-export const PROGRESS_SIMPLIFY_VALUES = ['applied', 'clean', 'reverted', 'not-run'];
+export const PROGRESS_SIMPLIFY_VALUES = ['applied', 'clean', 'reverted', 'revert-refused', 'not-run'];
 export const PROGRESS_REVIEW_GATE_EXITS = [0, 1, 2]; // K7: ALLOW/WARN, BLOCK, could not decide
 
 function checkStepFields(k, s, errors) {

lib/verification/criteria-runner.mjs

@@ -257,7 +257,13 @@ export function screenCommand(command, opts = {}) {
 //   - It is not a sandbox. `npm test`, `npm run <script>` and `make test` run
 //     whatever the repo's own package.json/Makefile says they run, including a
 //     script that pushes. That is the repo's responsibility, not this file's.
-//   - It has no opinion about arguments beyond the shape rules below.
+//   - It is not an option parser. Each runner takes only plain words (no
+//     quote, backslash, `$` or brace; a glob only inside a path) and a short
+//     list of options that only read (RUNNER_ARGS); an option that writes a
+//     file or loads code (`--test-reporter-destination`, `--outputFile`,
+//     `--junitxml`, `--basetemp`, `--import`, `--config`, `-o`, `-exec`) is on
+//     no list (release review of 6.3.0, round 3: rounds 1-2 closed this for
+//     git only).
 //   - It rejects plenty of honest commands: an env prefix (`CI=1 npm test`), a
 //     project's own binary (`./build/app --check`), anything piped. A check
 //     that needs one of those is declared through a runner instead —
@@ -279,6 +285,60 @@ function writesOutput(word) {
 const deny = (reason) => ({ allowed: false, reason });
 const permit = () => ({ allowed: true, reason: '' });
 
+// A word /bin/sh hands on unchanged, or a path with a glob in it: a glob can
+// only expand to names that start with the word's first character, so it
+// cannot make an option out of a path.
+const PLAIN_WORD = /^(?:[A-Za-z0-9_@%^~:.,=+/-]+|[A-Za-z0-9_./][A-Za-z0-9_@%^~:.,=+/*?-]*)$/;
+
+// After each runner's fixed words: the options it may take (`name` or
+// `name=value`; `value` lists the ones whose value may be the next word, and
+// `only` the ones whose value must be one of a few words: a built-in node test
+// reporter only formats, a reporter path would load code),
+// whether `--` may hand the rest to the repo's own script (`dashDash`: then
+// only paths follow), and whether a plain word may carry `=` (make reads
+// `VAR=value` as an override). The lists hold options that only read.
+const RUNNER_ARGS = {
+  npm: { fixed: (t) => (t[1] === 'run' ? 3 : 2), opts: [], dashDash: true },
+  node: {
+    fixed: 2,
+    opts: ['--test-name-pattern', '--test-skip-pattern', '--test-only', '--test-concurrency', '--test-timeout', '--test-reporter'],
+    value: ['--test-reporter'],
+    only: { '--test-reporter': ['spec', 'tap', 'dot', 'junit', 'lcov'] },
+  },
+  vitest: { fixed: 1, opts: ['--run', '--passWithNoTests', '--silent', '--bail'] },
+  jest: { fixed: 1, opts: ['--ci', '--runInBand', '-i', '--silent', '--passWithNoTests', '--bail', '--verbose'] },
+  pytest: { fixed: 1, opts: ['-q', '-qq', '-v', '-vv', '-x', '-s', '--quiet', '--verbose', '--exitfirst', '--tb', '--maxfail'] },
+  cargo: { fixed: 2, opts: ['--release', '--workspace', '--all', '--lib', '--doc', '-q', '--quiet', '--no-fail-fast'] },
+  go: { fixed: 2, opts: ['-v', '-short', '-race', '-failfast', '-cover', '-count', '-run'], value: ['-count', '-run'] },
+  make: { fixed: 2, opts: [], noAssign: true },
+};
+RUNNER_ARGS.python = RUNNER_ARGS.python3 = RUNNER_ARGS.uv = { ...RUNNER_ARGS.pytest, fixed: 3 };
+
+function screenArgs(t, spec) {
+  const fixed = typeof spec.fixed === 'function' ? spec.fixed(t) : spec.fixed;
+  for (let i = fixed, paths = false; i < t.length; i++) {
+    const w = t[i];
+    if (paths) {
+      if (w.startsWith('-')) return deny(`an option after \`--\` (${w}): only paths may follow it`);
+      continue;
+    }
+    if (w === '--' && spec.dashDash) { paths = true; continue; }
+    if (w.startsWith('-')) {
+      const name = w.split('=')[0];
+      if (!spec.opts.includes(name)) return deny(`\`${t[0]}\` with an option that is not on its read-only list (${name})`);
+      const inline = w.includes('=');
+      const value = inline ? w.slice(name.length + 1) : spec.value?.includes(name) ? t[i + 1] : undefined;
+      if (!inline && spec.value?.includes(name)) i++;
+      if (spec.only?.[name] && !spec.only[name].includes(value)) {
+        return deny(`\`${t[0]}\` with ${name} ${value ?? '(no value)'}: only ${spec.only[name].join('|')}`);
+      }
+      continue;
+    }
+    if (spec.noAssign && w.includes('=')) return deny(`\`${t[0]}\` with a variable override (${w})`);
+  }
+  return permit();
+}
+
 // `bash <path>` is the way a repo declares a check the allowlist has no name
 // for. The path must live under tests/, be relative, and not climb out of it.
 function isTestScript(path) {
@@ -371,7 +431,11 @@ function screenShape(command, opts, writeTime) {
 
   const tokens = text.trim().split(/\s+/);
   if (!Object.hasOwn(RUNNERS, tokens[0])) return deny(`\`${tokens[0]}\` is not a known test runner`);
-  return RUNNERS[tokens[0]](tokens, cwd, writeTime);
+  const form = RUNNERS[tokens[0]](tokens, cwd, writeTime);
+  if (!form.allowed || tokens[0] === 'git') return form;
+  const odd = tokens.find((w) => !PLAIN_WORD.test(w));
+  if (odd !== undefined) return deny(`a word the shell would rewrite (${odd}): only plain words are checked`);
+  return Object.hasOwn(RUNNER_ARGS, tokens[0]) ? screenArgs(tokens, RUNNER_ARGS[tokens[0]]) : permit();
 }
 
 // --- running ----------------------------------------------------------------

lib/verification/plan-scope-sweep.mjs

@@ -33,7 +33,7 @@ import { readFileSync, existsSync, realpathSync } from 'node:fs';
 import { spawnSync } from 'node:child_process';
 import { resolve, relative, dirname, isAbsolute, sep } from 'node:path';
 import { fileURLToPath } from 'node:url';
-import { sliceSteps, parseStepFields } from '../parsers/plan-schema.mjs';
+import { sliceSteps, parseStepFields, fieldPaths } from '../parsers/plan-schema.mjs';
 import { parseManifest } from '../parsers/manifest-yaml.mjs';
 
 // Files Voyage's own process writes while planning and executing.
@@ -44,9 +44,8 @@ export const PROCESS_FILES = Object.freeze([
 ]);
 
 const BACKTICK_SPAN = /`([^`\n]+)`/g;
-// Backticked: spaces allowed (`docs/my file.md`); bare: no whitespace.
+// A backticked path in `Test first:` (spaces allowed: `docs/my file.md`).
 const looksLikeTickedPath = s => /[./]/.test(s) && !/^https?:/.test(s) && !s.startsWith('-') && s.trim() === s;
-const looksLikePath = s => !/\s/.test(s) && looksLikeTickedPath(s);
 
 function pathsIn(text) {
   const out = [];
@@ -62,12 +61,9 @@ export function plannedPaths(planText) {
   for (const s of sliceSteps(planText)) {
     const f = parseStepFields(s.body);
     const files = f.get('Files');
-    if (files) {
+    // One reader with the tests-only check (plan-schema.mjs fieldPaths):
-      const ticked = pathsIn(files.text);
+    // backticked spans as written, else every path-like word.
-      // Older plans write bare paths: `- Files: a.ts, b.ts`.
+    if (files) for (const p of fieldPaths(files.text)) out.add(norm(p));
-      const bare = ticked.length ? [] : files.text.split(/[,\s]+/).filter(looksLikePath);
-      for (const p of [...ticked, ...bare]) out.add(norm(p));
-    }
     const tf = f.get('Test first');
     if (tf) {
       for (const line of tf.text.split('\n')) {

scripts/release-check.mjs

@@ -39,7 +39,6 @@ import { dirname, join, relative, resolve } from 'node:path';
 import { fileURLToPath } from 'node:url';
 import { decideVerdict } from '../lib/review/verdict-gate.mjs';
 import { commandEnv } from '../lib/util/command-env.mjs';
-import { parseDocument } from '../lib/util/frontmatter.mjs';
 
 const PLUGIN_ROOT = resolve(dirname(fileURLToPath(import.meta.url)), '..');
 
@@ -54,6 +53,11 @@ export const CHECK_IDS = Object.freeze(['tree', 'suite', 'review', 'coverage', '
 export const RELEASE_BOOKKEEPING = Object.freeze([
   'CHANGELOG.md', 'package.json', 'package-lock.json', '.claude-plugin/plugin.json', 'README.md',
 ]);
+// Voyage's own pipeline artifacts, committed by a repo that tracks them
+// (`artifact_store: git`, `.claude/voyage/approvals.jsonl`): the review itself
+// and the approval log land after the reviewed commit by design (release
+// review of 6.3.0, round 3). The plan-file sweep treats them the same way.
+const PROCESS_PREFIXES = Object.freeze(['.claude/projects/', '.claude/voyage/']);
 const BADGE_RE = /badge\/version-/;
 const VERSION_FILES = Object.freeze(['package.json', 'package-lock.json', '.claude-plugin/plugin.json']);
 
@@ -165,11 +169,20 @@ function checkReview(reviewPath) {
   return { check: { id: 'review', status: 'pass', detail: `review verdict ${v.verdict}` }, text };
 }
 
+/**
+ * The review's `scope_sha_end` as written. Read from the frontmatter text, not
+ * through the YAML reader, which turns an all-digit value into a number and
+ * drops a leading zero (`0123456` → 123456; release review of 6.3.0, round 3).
+ */
+export function reviewedSha(reviewText) {
+  const fm = /^---\r?\n([\s\S]*?)\r?\n---\r?\n/.exec(String(reviewText ?? ''));
+  const m = fm && /^scope_sha_end:[ \t]*(["']?)([^"'\s#]*)\1[ \t]*(?:#.*)?$/m.exec(fm[1]);
+  return m ? m[2] : '';
+}
+
 function checkCoverage(repo, reviewText) {
   if (reviewText === null) return { id: 'coverage', status: 'skip', detail: 'not checked: no readable review' };
-  let sha;
+  const sha = reviewedSha(reviewText);
-  try { sha = parseDocument(reviewText).parsed?.frontmatter?.scope_sha_end; } catch { sha = undefined; }
-  sha = sha === undefined || sha === null ? '' : String(sha).trim();
   if (!sha) return { id: 'coverage', status: 'fail', detail: 'review.md has no scope_sha_end — cannot tell what it reviewed' };
   const resolved = git(repo, ['rev-parse', '--verify', '--quiet', `${sha}^{commit}`]);
   if (resolved.status !== 0) return { id: 'coverage', status: 'fail', detail: `the reviewed commit ${sha} is not in the history of HEAD (unknown commit)` };
@@ -182,7 +195,7 @@ function checkCoverage(repo, reviewText) {
   const names = git(repo, ['diff', '--relative', '--name-only', full, 'HEAD']);
   if (names.status !== 0) return { id: 'coverage', status: 'error', detail: `git diff failed: ${names.stderr.trim()}` };
   const changed = names.stdout.split('\n').filter(Boolean);
-  const outside = changed.filter((f) => !RELEASE_BOOKKEEPING.includes(f));
+  const outside = changed.filter((f) => !RELEASE_BOOKKEEPING.includes(f) && !PROCESS_PREFIXES.some((p) => f.startsWith(p)));
   for (const file of changed.filter((f) => VERSION_FILES.includes(f))) {
     const before = git(repo, ['show', `${full}:./${file}`]);
     const after = git(repo, ['show', `HEAD:./${file}`]);

tests/fixtures/jsonl-schemas.md

@@ -22,7 +22,7 @@
 | trekbrief-stats | ts, task, slug, mode, interview_turns, review_iterations, brief_quality, research_topics, auto_research, auto_result, intent_approved, project_dir, order_id, order_claimed_at, order_reason | commands/trekbrief.md (orchestrator-emit Phase 7; order_* added by lib/stats/stats-append.mjs) | trekbrief.md:1130-1145 | profile, phase_models, profile_source | none |
 | trekresearch-stats | ts, question, mode, scope, engine, slug, project_dir, brief_path, dimensions, dimensions_baseline, dimensions_baseline_preserved, effort, conv_turns, empty_turns, unique_sources, agents_local, agents_external, gemini_used, confidence, contradictions, open_questions | lib/stats/stats-append.mjs (trekresearch.md Stats tracking) | trekresearch.md:782-808 | profile, phase_models, parallel_agents, external_research_enabled, profile_source | none |
 | trekplan-stats | ts, task, mode, slug, brief_path, project_dir, codebase_size, codebase_files, agents_deployed, deep_dives, research_briefs_used, research_scout_used, critic_verdict, guardian_verdict, outcome | lib/stats/stats-append.mjs (trekplan.md Phase 12) | trekplan.md:1115-1135 | profile, phase_models, parallel_agents, profile_source | none |
-| trekexecute-stats (Phase 9 record) | ts, plan, plan_type, mode, result, steps_total, steps_passed, steps_failed, steps_skipped, failed_at_step | lib/stats/stats-append.mjs (trekexecute.md Phase 9) | trekexecute.md:2877-2899 | profile, phase_models, profile_source | none |
+| trekexecute-stats (Phase 9 record) | ts, plan, plan_type, mode, result, steps_total, steps_passed, steps_failed, steps_skipped, failed_at_step | lib/stats/stats-append.mjs (trekexecute.md Phase 9) | trekexecute.md:2891-2913 | profile, phase_models, profile_source | none |
 | trekexecute-stats (autonomy events) | ts, event, known_event, payload | lib/stats/event-emit.mjs `emit()` | event-emit.mjs:78-99 | payload.profile, payload.phase_models, payload.profile_source | none |
 | trekexecute-stats (PostToolUse Bash) | ts, session_id, command_excerpt, duration_ms, success | hooks/scripts/post-bash-stats.mjs (Bash PostToolUse) | post-bash-stats.mjs:42-54 | none (hook is plugin-level, not profile-aware) | command_excerpt (CWE-212) |
 | trekreview-stats | ts, slug, verdict, counts (BLOCKER/MAJOR/MINOR/SUGGESTION), reviewed_files_count, mode, duration_ms | lib/stats/stats-append.mjs (trekreview.md Phase 8) | trekreview.md:628-632 | profile, phase_models, profile_source | none |

tests/lib/review-630.test.mjs

@@ -265,3 +265,116 @@ test('review 630 r2 (operator 2026-10-10): Phase 7.2 says plainly that template-
   assert.match(cmd, /Today the loop runs only for steps written as\s+runnable plan steps/);
   assert.match(cmd, /re-plan from the review" — the\s+usual outcome until a later order/);
 });
+
+// ==== round 3 of the release review (2026-10-10), each red on 96cb977 ===========
+
+test('review 630 r3 / BLOCKER: no runner takes an option that writes a file or loads code', async () => {
+  const { allowedCommand } = await import('../../lib/verification/criteria-runner.mjs');
+  for (const cmd of [
+    'node --test --test-reporter=tap --test-reporter-destination=/tmp/x',
+    'node --test --import=data:text/javascript,1',
+    'node --test "--import=data:text/javascript,import(\'node:fs\')"',
+    'node --test --require ./evil.cjs',
+    'node --test --test-reporter=./evil-reporter.mjs',
+    'node --test --test-reporter ./evil-reporter.mjs',
+    'jest --json --outputFile=/tmp/x',
+    'jest --config=evil.js',
+    'vitest --outputFile=/tmp/x',
+    'vitest run --reporter=./evil.mjs',
+    'pytest --junitxml=/tmp/x',
+    'pytest --basetemp=/tmp/anydir',
+    'pytest -p evil_plugin',
+    'python -m pytest --junitxml=/tmp/x',
+    'uv run pytest --basetemp=/tmp/x',
+    'go test -o /tmp/x',
+    'go test -exec /tmp/evil ./...',
+    'cargo test --target-dir=/tmp/x',
+    'npm test -- --test-reporter-destination=/tmp/x',
+    'npm test --script-shell=/tmp/evil',
+    'npm run verify -- --outputFile=/tmp/x',
+    'make test SHELL=/tmp/evil',
+    'make test -f /tmp/evil.mk',
+    'bash tests/x.sh "$(id)"',
+  ]) {
+    assert.equal(allowedCommand(cmd, { cwd: ROOT }).allowed, false, `${cmd} must not reach the shell`);
+  }
+  for (const cmd of [
+    'npm test', 'npm test -- tests/lib/criteria-runner.test.mjs', 'node --test tests/lib/criteria-runner.test.mjs',
+    'node --test tests/lib/*.test.mjs', 'node --test --test-reporter=tap tests/lib/review-630.test.mjs', 'node --test --test-reporter dot', 'node --test --test-name-pattern=review tests/lib/review-630.test.mjs',
+    'vitest run', 'jest --ci', 'pytest -q', 'pytest -x -q tests/', 'python3 -m pytest tests/', 'uv run pytest',
+    'cargo test', 'cargo test --release', 'go test ./...', 'go test -run TestX -count=1 ./pkg', 'make test',
+    'bash tests/fixtures/criteria-exit-0.sh', 'bash tests/fixtures/criteria-exit-0.sh --fast',
+  ]) {
+    const r = allowedCommand(cmd, { cwd: ROOT });
+    assert.equal(r.allowed, true, `the control: ${cmd} is a test run (${r.reason})`);
+  }
+});
+
+test('review 630 r3: a restore the guard refuses is never recorded as reverted', async () => {
+  const { PROGRESS_SIMPLIFY_VALUES } = await import('../../lib/validators/progress-validator.mjs');
+  assert.ok(PROGRESS_SIMPLIFY_VALUES.includes('revert-refused'), 'progress.json can say the restore was refused');
+  assert.match(read('docs/HANDOVER-CONTRACTS.md'), /`applied \\| clean \\| reverted \\| revert-refused \\| not-run`/);
+  const cmd = read('commands/trekexecute.md');
+  const onFailure = cmd.slice(cmd.indexOf('**`On failure: revert`** and **`On failure: retry`**'), cmd.indexOf('**`On failure: skip`**'));
+  assert.match(onFailure, /guard refuses the restore/, 'the step revert says what happens when the guard refuses it');
+  const simplify = cmd.slice(cmd.indexOf('2. **`/simplify`**'), cmd.indexOf('3. **`/trekreview`**'));
+  assert.match(simplify, /simplify = "revert-refused"/);
+  assert.match(simplify, /never `"reverted"`/);
+});
+
+test('review 630 r3: a fixture document under tests/ is a test file, and "e.g." is not a path', () => {
+  const step = (files) => `- **Files:** ${files}\n- **Changes:** x\n`;
+  assert.equal(stepTestsOnly(step('`tests/a.test.mjs`, `tests/fixtures/sample.md`')).testsOnly, true);
+  assert.deepEqual(stepTestsOnly(step('tests/a.test.mjs (new), e.g. the fixture tests/fixtures/x.json')),
+    { testsOnly: true, files: ['tests/a.test.mjs', 'tests/fixtures/x.json'] });
+  assert.equal(stepTestsOnly(step('`docs/specs/api.md`')).testsOnly, false, 'a document outside tests/ is still not a test');
+  assert.deepEqual(stepTestsOnly(step('`./tests/a.test.mjs`')).files, ['tests/a.test.mjs'], './ is stripped');
+});
+
+test('review 630 r3: the plan-file sweep reads Files: with the same reader as the tests-only check', async () => {
+  const { plannedPaths } = await import('../../lib/verification/plan-scope-sweep.mjs');
+  const plan = '## Implementation Plan\n\n### Step 1: x\n\n- **Files:** `./src/a.mjs`, tests/b.test.mjs + src/c.mjs\n- **Changes:** x\n';
+  const body = plan.slice(plan.indexOf('### Step 1'));
+  assert.deepEqual(plannedPaths(plan), [...stepTestsOnly(body).files].sort());
+});
+
+test('review 630 r3: --steps-from refuses an appended step with no Verify: on a 1.8 plan too', () => {
+  const plan = buildPlan({});
+  const step = '### Step 3: Fix it\n\n- **Files:** `src/fix.mjs`, `tests/fix.test.mjs`\n- **Changes:** fix it\n' +
+    '- **Test first:**\n  - File: `tests/fix.test.mjs` (new)\n  - Verifies: it is fixed\n- **On failure:** escalate\n';
+  const edited = plan.replace('\n## Test Strategy', `\n${step}\n## Test Strategy`);
+  assert.ok(edited.includes('### Step 3: Fix it'), 'the fixture appends the step');
+  const r = validatePlanContent(edited, { strict: true, stepsFrom: 3 });
+  assert.ok(r.errors.some((e) => e.code === 'PLAN_STEP_VERIFY_NOT_RUNNABLE' && /Step 3/.test(e.message)),
+    JSON.stringify(r.errors.map((e) => e.code)));
+});
+
+test('review 630 r3: the plan validator hands its parsed fields to the tests-only check', async () => {
+  const schema = await import('../../lib/parsers/plan-schema.mjs');
+  assert.equal(typeof schema.stepTestsOnlyFromFields, 'function');
+  const body = '- **Files:** `tests/a.test.mjs`\n- **Changes:** x\n';
+  assert.deepEqual(schema.stepTestsOnlyFromFields(schema.parseStepFields(body)), stepTestsOnly(body));
+  assert.match(read('lib/validators/plan-validator.mjs'), /stepTestsOnlyFromFields\(f\)/);
+});
+
+test('review 630 r3: skill frontmatter - a value starting with # is a hazard, an escaped backslash closes a quote', async () => {
+  const { plainScalarHazards, readFrontmatterFields } = await import('../../lib/skills/skill-frontmatter.mjs');
+  assert.ok(plainScalarHazards('#1 helper. Use when testing.').length > 0, 'YAML reads "#..." as a comment: the value is null');
+  const r = readFrontmatterFields('---\nname: x\ndescription: "Path C:\\\\"\n---\nbody\n');
+  assert.equal(r.fields.description?.value, 'Path C:\\', JSON.stringify(r));
+  const open = readFrontmatterFields('---\nname: x\ndescription: "a \\"\n---\nbody\n');
+  assert.ok(JSON.stringify(open).includes('unclosed double quote'), 'a quote escaped by one backslash is still open');
+});
+
+test('review 630 r3: routing reads a kind tag only from its own table, never Object.prototype', () => {
+  const dir = mkdtempSync(join(tmpdir(), 'r630-proto-'));
+  try {
+    for (const d of ['skills', 'commands', 'agents', 'evals']) cpSync(join(ROOT, d), join(dir, d), { recursive: true });
+    const c = join(dir, 'evals', 'zz-proto');
+    cpSync(join(ROOT, 'evals', 'skill-test-first-quiet-1'), c, { recursive: true });
+    writeFileSync(join(c, 'prompt.md'), '---\ntags: [constructor, owner=nobody]\n---\n\nWhat time is it?\n');
+    const { checkRouting } = awaitImportSync();
+    const r = checkRouting(dir);
+    assert.ok(!r.errors.some((e) => String(e.message ?? '').includes('zz-proto') || e.case === 'zz-proto'), JSON.stringify(r.errors));
+  } finally { rmSync(dir, { recursive: true, force: true }); }
+});

tests/scripts/release-check.test.mjs

@@ -126,7 +126,10 @@ function fixture({
   git(dir, 'commit', '-q', '-m', 'chore(release): fixture');
 
   if (afterRelease) {
-    for (const [rel, text] of Object.entries(afterRelease)) writeFileSync(join(dir, rel), text);
+    for (const [rel, text] of Object.entries(afterRelease)) {
+      mkdirSync(dirname(join(dir, rel)), { recursive: true });
+      writeFileSync(join(dir, rel), text);
+    }
     git(dir, 'add', '-A');
     git(dir, 'commit', '-q', '-m', 'after the review');
   }
@@ -388,3 +391,23 @@ test('release-check — a flag with no value is exit 2, not 1', () => {
   assert.equal(r.status, 2, r.stdout + r.stderr);
   assert.match(r.stdout + r.stderr, /not measured|NOT MEASURED/i);
 });
+
+// ==== round 3 of the release review of 6.3.0 (2026-10-10) ====================
+
+test('release-check r3 — the review and the approval log committed after the review do not count as unreviewed → 0', () => {
+  const fx = fixture({ afterRelease: {
+    '.claude/projects/2026-10-08-s/review.md': 'the review, committed (artifact_store: git)\n',
+    '.claude/voyage/approvals.jsonl': '{"kind":"release"}\n',
+  } });
+  try {
+    const r = run(fx);
+    assert.equal(r.status, 0, r.stdout + r.stderr);
+  } finally { fx.cleanup(); }
+});
+
+test('release-check r3 — scope_sha_end is read as written: an all-digit sha keeps its leading zero', () => {
+  assert.equal(typeof mod.reviewedSha, 'function');
+  assert.equal(mod.reviewedSha('---\ntype: trekreview\nscope_sha_end: 0123456\n---\nbody\n'), '0123456');
+  assert.equal(mod.reviewedSha('---\nscope_sha_end: "00ab12"\n---\n'), '00ab12');
+  assert.equal(mod.reviewedSha('---\ntype: trekreview\n---\nscope_sha_end: 1\n'), '', 'only the frontmatter counts');
+});