Files
msd-core/src/check-command-router.cts
Behruz Nassre Esfahani d04e287fa9 fix(#2365): stop api-coverage detector false-positiving non-API phases (#2397)
* fix(#2365): stop the api-coverage detector false-positiving non-API phases

detectApiIntegration fired on any integration verb co-occurring anywhere on a
line with any API noun, treated / as a word boundary (so a first-party Next.js
src/app/api/... route path matched the noun "api"), and read any capitalized
word before API/SDK/REST/GraphQL as a service name behind a fixed stopword
denylist (so threat-model prose like "Resolver-only API" fired). Because the
verify:pre seal gate is BLOCKING, a phase touching no external API could not
reach UAT without fabricating a coverage matrix.

The compound rule now requires the verb and noun to share one clause (sentence
punctuation and table-cell walls end a clause) within a bounded word gap.
Non-prose spans are excluded before matching: fenced code (already), inline
code spans (new stripInlineCode in the markdown-sectionizer seam), and
path-shaped tokens. The <Service> API surface rule requires proper-noun
position — a clause-initial capitalized word is ordinary English and needs
dependency evidence (URL / package reference) on the same line — and rejects
compound modifiers ("Resolver-only", lowercase after the hyphen).

A phase that integrates no external API now has a first-class, reasoned way to
say so: a COVERAGE.md containing "No external API integration: <reason>"
satisfies the gate (declaration + rows is contradictory and blocks). The
true-positive path is pinned by regression tests: every default-vocabulary
positive still fires, including the widest word-gap pairing and the
surface-rule-only shape.

Fixes #2365

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(#2365): tighten api-coverage detector per Codex review (round 2)

Applies the Codex review findings on the initial #2365 fix:

- S-1: a COVERAGE.md "no external API integration" declaration is the human
  override for a fallible detector, so it must PASS even when detection still
  fires — but the contradiction is now SURFACED in the gate output (overridden
  signal count + terms) instead of passing silently.
- S-2: verb/noun pairing is now a term-group nearest-pair merge walk over
  precomputed word ordinals (computeWordStarts / minWordGap), not a match×match
  cross product — a hostile line repeating one pair thousands of times stays
  linear instead of going quadratic.
- FN-4: package-shaped inline-code spans (`stripe-sdk`, `@stripe/stripe-js`)
  are kept as noun/dependency evidence rather than being fully masked, so a
  genuine dependency reference inside code ticks still corroborates.
- C-1: the <Service> API surface rule now scans every candidate in every
  clause; a rejected first candidate no longer shadows a later genuine service.
- Cross-clause binding: a verb may bind a noun in the immediately following
  clause only when its own clause names a service object, within a tight gap —
  admits "Integrate Stripe, exposing its endpoints …" without re-admitting the
  unrelated-clauses false-positive class.
- Internal-descriptor negative evidence ("internal Payments API",
  "the internal endpoint") never pairs; URL/scheme matching generalized beyond
  http(s).

All 5 acceptance criteria still hold: the three reported false positives are
clean and "integrate the Stripe API" still fires. Built .cjs committed
alongside the .cts. tsc + eslint (incl. no-adhoc-markdown-parsing) +
lint:regression-names clean; affected suites 256/256 green.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* fix(#2365): retune api-coverage detector fail-closed per Codex review (round 3)

Codex's second-round review found the round-2 tightening had over-corrected into
FAIL-OPEN false negatives — realistic external-API prose that the BLOCKING seal
gate silently let through (the catastrophic class, since a missed API surface is
worse than a dismissable false positive). Retuned the detector to be explicitly
fail-closed: lean toward detecting, and let the one-line COVERAGE.md "no external
API integration" declaration dismiss the residual false positives.

Fail-open false negatives fixed (all now detect):
- F1 clause-initial `<Service> API` with a plain follower ("Stripe API for
  payment processing") — dropped the follower-allowlist / corroboration gate on
  clause-initial surfaces; a service that is not a stopword, descriptor, or
  compound modifier is a real name from any clause position.
- F2 scheme-less external host ("api.stripe.com/v1") — a dotted host with an
  alphabetic final label now contributes its API nouns; a first-party route
  path (no dotted host) still does not.
- F3 vendor's first-party SDK ("Integrate Shopify's first-party SDK") — the
  compound path no longer filters nouns on "internal"/"first-party" (Codex: the
  qualifier can describe the vendor's own API, not the consuming project's).
- F4 long single integration clause — removed the word-gap cap entirely: it
  could not separate a 21-word genuine clause from an 18-word internal one, so
  the clause boundary is now the whole relationship test.
- F5 lowercase cross-clause service — cross-clause binding no longer requires a
  capitalized "service object".

New false positives fixed (all now clean):
- F6 a URL token that swallowed a trailing clause comma, merging two clauses —
  trailing clause punctuation is kept literal so the split survives.
- F7 a capitalized internal component authorizing cross-clause binding — the new
  gate requires a dependent elaboration, not a new coordinate clause opened by a
  conjunction ("…, then document…").
- F8 a protocol name read as a service ("REST API", "GraphQL API") — protocol
  and locality descriptors are rejected in the `<Service>` position.

- Finding 9: the inline-code-span scanner was O(n^2) on pathological backtick
  runs; rewritten to linear via a per-length run cursor (2 MB: 4.15 s -> ~6 ms),
  semantics preserved (148 sectionizer tests unchanged).

Net simplification: the fail-closed model removed the round-2 minWordGap /
groupByTerm / follower / corroboration machinery (350 insertions vs 445
deletions across the touched files). Under fail-closed, three round-2 negative
tests now correctly detect (integration verb + "internal"-qualified noun, and
the distant-same-clause case); none were trek-e acceptance FPs.

Verified: 1491/1491 unit tests pass; tsc + eslint (incl. no-adhoc-markdown-
parsing) + lint:regression-names clean; all 8 review findings reproduced as
regression tests, both directions. Built .cjs committed alongside the .cts.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* fix(#2365): resolve round-3 Codex review findings (fail-closed, round 4)

Codex's round-3 adversarial review found the fail-closed retune had introduced
new holes in both directions. Resolved:

Fail-open false negatives (now detect):
- External host addressing a PATH ("graph.microsoft.com/v1.0/me") is itself an
  integration surface and contributes an endpoint noun even when the host names
  no vocabulary word. A bare domain link with no path ("https://example.com")
  stays a non-signal, so "Integrate … from example.com, document …" is still
  clean.
- Locality qualification ("internal", "private") no longer leaks across a
  sentence or clause boundary: only plain spaces may separate the descriptor
  from the service, so "The cache is private. Stripe API …" now detects.
- Cross-clause binding: the fragile head-word cap (which could not tell a
  genuine "Connect … to Stripe payments, exposing its endpoints" from an
  unrelated "Integrate … from URL, document …" — both 4 words after the verb)
  is replaced by a participial-continuation rule: a verb binds a noun in the
  next clause only when that clause begins with an "-ing" elaboration. This
  fixes the 4-word-head false negative AND the false positive below at once.

False positives (now clean):
- Cross-clause no longer binds a finite continuation regardless of separator:
  "Wire the settings form. Document endpoint props." / "…; document …" /
  "…, document …" are separate actions, not elaborations.

Perf (quadratic → linear):
- The trailing-punctuation peel is a backward char scan instead of an
  unanchored `[…]+$` regex (16k chars: 156 ms → ~1 ms).
- SERVICE_SURFACE_API_RE bounds the service-name length {1,40} so a hostile
  "A-A-…-x" run cannot drive O(n^2) backtracking (16k: 385 ms → ~3 ms).

Consumer fail-open (blocking gate):
- readPhaseScope now distinguishes "no plans" from a plan that EXISTS but is
  unreadable. On a read error the gate BLOCKS ("could not read the phase
  scope …") instead of silently certifying no-integration from partial scope —
  an unreadable plan could be the one describing the integration.

Documented fail-closed tradeoffs, now pinned with tests so they are not
"fixed" back into a fail-open: a clause-initial capitalized common word before
"API" ("Payment API", "Search API") reads as a service name; a long clause
pairs a verb with a distant noun; and a CommonMark inline code span that wraps
a newline is matched within-line only. Codex judged these acceptable because
the COVERAGE.md declaration is a cheap override.

One documented limitation remains out of scope: "Integrate Stripe, and
authenticate requests with its API" (a coordinate finite clause whose noun
refers back by pronoun) needs coreference resolution, beyond a lexical detector.

Verified: 379/379 affected + command-router tests pass (+14 new regression
tests covering every round-3 finding, both directions); tsc + eslint
(no-adhoc-markdown-parsing) + lint:regression-names clean. Built .cjs committed.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* fix(#2365): simplify to robust core — remove whack-a-mole heuristics (round 5)

Round-4 review confirmed the detector's two most complex features generate
findings in both directions no matter how they are tuned, because they need a
vendor dictionary + coreference the issue rules out in principle. Per the
operator's "ship the robust core" decision, both are removed and their gaps are
documented rather than chased further:

- Cross-clause binding DELETED (allowsCrossClause / participle rule). It caused
  a fail-open on finite continuations ("Integrate Stripe; use its OAuth
  endpoints" — missed) and a false positive on "-ing"-SPELLED nouns ("…, billing
  endpoint terminology…" — wrongly fired). Detection is now same-clause only.
- URL-path-as-evidence REVERTED. Treating every path-bearing URL as an endpoint
  fired on ordinary asset/link URLs ("…/theme.css", "…?next=/x", a docs/repo
  link) and recreated routine UI-phase false positives. An external URL is
  evidence only when it NAMES an API vocabulary word ("api.stripe.com/v1").

Two fail-open cases are now DOCUMENTED limitations, pinned by tests so a future
maintainer does not re-add the heuristics that caused the false positives above:
a service named only in a clause separate from its API noun, and a bare external
host that names no vocabulary word. Both are cheaply covered by the COVERAGE.md
declaration and rare in real phase prose ("integrate the X API").

Also fixed from the round-4 review:
- Qualification now survives markdown emphasis ("The **internal** Payments API"
  stays clean) while still not crossing a sentence/clause boundary.
- readPhaseScope fail-closes on a REAL read failure (EACCES/EIO) enumerating the
  phase directory or reading the roadmap fallback — not only per-plan-file
  failures; a missing directory/section remains a legitimate no-op. The
  declaration-override path surfaces scope_read_error so an incomplete-scope
  override stays visible.
- SERVICE_SURFACE_API_RE length-bound comment no longer overclaims.

Net: the detector is same-clause verb+noun + `<Service> API` surface, with
path/code/inline masking and a fail-closed posture. All five acceptance criteria
hold. 1573/1573 unit tests pass; tsc + eslint (no-adhoc-markdown-parsing) +
lint:regression-names clean. Built .cjs committed.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* fix(#2365): close roadmap-fallback fail-open + stale JSDoc (round-5 review)

The round-5 sanity review confirmed the detector simplification is sound (all
acceptance positives fire, all required negatives clean) and flagged one real
blocker plus a nit:

- Blocker: readPhaseScope's roadmap fallback could still silently pass an
  UNREADABLE roadmap. getRoadmapPhaseWithFallback gated on fs.existsSync(), which
  returns false on EACCES/EIO too — so an unreadable ROADMAP.md read as "absent",
  no exception reached isRealReadFailure, and the blocking gate certified empty
  scope. Fixed at the source: read the roadmap directly and honor the function's
  OWN documented contract — null only on ENOENT (genuinely absent), otherwise
  throw. Both existing callers already wrap it in try/catch expecting that throw,
  and readPhaseScope now fail-closes (blocks) via its roadmap catch. Verified by
  a new e2e test (unreadable roadmap fallback → block).

- Nit: the detectApiIntegration JSDoc still described the removed cross-clause
  participial binding and "every external hostname counts" — corrected to the
  actual same-clause-only behavior and the names-a-vocab-word URL rule.

Verified: full unit suite green; tsc + eslint + lint:regression-names clean.
Built .cjs committed (roadmap.cjs is gitignored/rebuilt, per repo convention).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* chore(#2365): backfill changeset PR number (#2397)

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* chore(#2365): sync generated capability-registry + recapture install goldens

CI surfaced two generated-artifact staleness issues (all failing test shards +
lint-tests traced to these, not to a logic defect):

- gsd-core/bin/lib/capability-registry.cjs was stale: the initial fix edited the
  ai-integration `api-coverage-plan-pre.md` fragment (added the "No external API
  integration" declaration section) but did not regenerate the registry, which
  embeds an inline copy of that fragment. Regenerated via
  `gen-capability-registry.cjs --write` — the diff is exactly the fragment text
  sync. Fixes `lint:generated-sync` and the "committed registry is in sync" +
  "registry integration" tests.

- The 18 golden-install-parity fixtures were stale by exactly one hash line each
  — `gsd-core/references/api-coverage.md`, which this PR edits and which is a
  hashed installed artifact. Recaptured with `UPDATE_GOLDEN=1`; the diff is that
  single hash per runtime and nothing else. Fixes the `golden parity — *` tests.

No source or behavior change — generated artifacts only.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* test(#2365): flip representative-corpus manifest to assert the fixed behavior

The #2371 representative corpus (merged into next after this branch was cut) is a
known-bug tripwire: it asserts each fixture's currentBuggyOutput so the test
fails loudly the moment #2365 is fixed, at which point — per its own contract in
representative-corpus.test.cjs — the fixer removes currentBuggyOutput so the
assertion checks expectedDetected instead.

This is that moment. Removed currentBuggyOutput from the three detector fixtures
(nextjs-route-path, unrelated-verb-noun, threat-model-prose); the corpus now
asserts detected:false, which the fail-closed same-clause detector satisfies.
Notes updated to describe the fix rather than the bug. The #2366 matrix corpus
is left untouched — that tripwire belongs to its own PR (#2374).

Corpus test: 7/7 pass.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* test(#2365): skip chmod-000 fail-closed e2e tests on Windows

The three fail-closed gate tests induce an unreadable plan / directory / roadmap
with chmod 000, but Windows does not enforce POSIX mode bits — readFileSync
still succeeds, so the gate never reaches the read-error path and the assertion
fails on the windows-latest CI leg. The fail-closed LOGIC is platform-
independent (readError → block) and is fully exercised on the macOS/Linux legs;
only the method of inducing EACCES is POSIX-specific. Guard the three tests to
skip on win32 as well as root, mirroring golden-install-parity's win32 skip.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* test(#2365): address trek-e review — glossary, clock-seam, IO injection, bounds

Review response to PR #2397 (trek-e, CHANGES_REQUESTED). Fix logic unchanged;
this closes the test/process-hygiene findings.

Major:
- CONTEXT.md "Markdown Sectionizer" glossary now lists the two exports this fix
  relies on, `stripInlineCode` and `scanInlineCodeSpans` (glossary is a PR gate).
- Replaced the banned wall-clock assertion in the "hostile repeated-term line"
  test (Clock Seams rule — no elapsed-time asserts) with a deterministic
  signal-count assertion, which also directly verifies the term-dedup that keeps
  pairing linear (one signal for a 10k-pair line, not thousands).
- Rewrote the three fail-closed read-failure tests: instead of chmod 0o000
  (a no-op under root / on Windows, the pattern the repo's IO-failure convention
  avoids) they now exercise the newly-exported `readPhaseScope` in-process and
  inject the failure by monkeypatching fs.readFileSync/readdirSync to throw,
  restoring in finally. Deterministic and platform-independent (no skip needed),
  and they add the ENOENT-is-absence case that the chmod tests couldn't express.

Minor:
- Added limit / limit+1 boundary tests for SERVICE_SURFACE_API_RE's {1,40}
  service-name bound, QUALIFIER_LOOKBACK's 24-char window, and REASON_MAX_LEN
  (200) on the declaration reason.
- Added a fast-check property that fuzzes the tokenizer / clause splitter /
  masking (scanLineTokens, splitClauses, collectTermMatches) with adversarial
  tokens (slashes, backticks, URLs, clause punctuation) and asserts the detector
  is total (never throws), shape-stable, holds detected <=> signals, and is
  deterministic.

readPhaseScope is exported for the in-process tests. Verified: 125 detector +
19 gate tests pass; tsc + eslint + generated-sync (glossary/registry) +
lint-regression-test-names + lint-test-file-count clean.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-07-19 14:59:25 -04:00

1457 lines
58 KiB
TypeScript

/**
* Check subcommand router — auto-mode, decision-coverage-plan, decision-coverage-verify.
*
* ADR-457 build-at-publish: the hand-written bin/lib/check-command-router.cjs collapsed
* to a TypeScript source of truth. Behaviour is preserved byte-for-behaviour
* from the prior hand-written .cjs; only strict types are added.
*/
import fs from 'node:fs';
import path from 'node:path';
import { execFileSync } from 'node:child_process';
// eslint-disable-next-line @typescript-eslint/no-require-imports
import io = require('./io.cjs');
const { output, error, ERROR_REASON } = io;
// eslint-disable-next-line @typescript-eslint/no-require-imports
import planningWorkspaceMod = require('./planning-workspace.cjs');
const { planningDir } = planningWorkspaceMod;
// eslint-disable-next-line @typescript-eslint/no-require-imports
import phaseLocatorMod = require('./phase-locator.cjs');
const { findPhaseInternal } = phaseLocatorMod;
import { extractDecisions } from './decisions.cjs';
import type { Decision } from './decisions.cjs';
import { stripFencedCode, collectSections } from './markdown-sectionizer.cjs';
import { checkUiPresence } from './ui-safety-gate.cjs';
// eslint-disable-next-line @typescript-eslint/no-require-imports
import verifyModule = require('./verify.cjs');
const { cmdVerifySchemaDrift, cmdVerifyCodebaseDrift } = verifyModule;
// eslint-disable-next-line @typescript-eslint/no-require-imports
import roadmapModule = require('./roadmap.cjs');
const { getRoadmapPhaseWithFallback } = roadmapModule;
// eslint-disable-next-line @typescript-eslint/no-require-imports
import gapCheckerModule = require('./gap-checker.cjs');
const { runGapAnalysis } = gapCheckerModule;
import { routeProhibitionEnforcement } from './prohibition-enforcement.cjs';
// eslint-disable-next-line @typescript-eslint/no-require-imports
import gatePredicateEval = require('./gate-predicate-evaluator.cjs');
const { evaluatePredicate } = gatePredicateEval;
// eslint-disable-next-line @typescript-eslint/no-require-imports
import apiCoverageMod = require('./api-coverage.cjs');
const { detectApiIntegration, validateCoverageMatrix } = apiCoverageMod;
import { execTool, posixNormalize } from './shell-command-projection.cjs';
// ─── Helpers ──────────────────────────────────────────────────────────────────
function normalizePhrase(text: unknown): string {
// eslint-disable-next-line @typescript-eslint/no-base-to-string
return String(text || '')
.toLowerCase()
.replace(/[^a-z0-9\s]/g, ' ')
.replace(/\s+/g, ' ')
.trim();
}
const SOFT_PHRASE_MIN_WORDS = 6;
function softPhrase(text: unknown): string {
const words = normalizePhrase(text).split(' ').filter(Boolean);
if (words.length < SOFT_PHRASE_MIN_WORDS) return '';
return words.slice(0, SOFT_PHRASE_MIN_WORDS).join(' ');
}
function decisionMentioned(haystack: string | null | undefined, decision: Decision): boolean {
if (!haystack) return false;
if (new RegExp(`\\b${decision.id}\\b`).test(haystack)) return true;
const phrase = softPhrase(decision.text);
return phrase ? normalizePhrase(haystack).includes(phrase) : false;
}
function readIfExists(filePath: string): string {
try {
return fs.readFileSync(filePath, 'utf-8');
} catch {
return '';
}
}
function resolvePath(inputPath: string, projectDir: string): string {
return path.isAbsolute(inputPath) ? inputPath : path.join(projectDir, inputPath);
}
interface WorkflowConfig {
auto_advance?: boolean;
_auto_chain_active?: boolean;
context_coverage_gate?: boolean | string;
}
function readWorkflowConfig(projectDir: string): WorkflowConfig {
const configPath = path.join(projectDir, '.planning', 'config.json');
try {
const parsed = JSON.parse(fs.readFileSync(configPath, 'utf-8')) as Record<string, unknown>;
const wf = (parsed['workflow'] as Record<string, unknown> | undefined) || {};
return {
...wf,
auto_advance: (wf['auto_advance'] ?? parsed['auto_advance']) as boolean | undefined,
_auto_chain_active: (wf['_auto_chain_active'] ?? parsed['_auto_chain_active']) as boolean | undefined,
context_coverage_gate: (wf['context_coverage_gate'] ?? parsed['context_coverage_gate']) as boolean | string | undefined,
};
} catch {
return {};
}
}
function cmdAutoMode(projectDir: string, raw: boolean): void {
const workflow = readWorkflowConfig(projectDir);
const autoAdvance = Boolean(workflow.auto_advance ?? false);
const autoChainActive = Boolean(workflow._auto_chain_active ?? false);
let source = 'none';
if (autoChainActive && autoAdvance) source = 'both';
else if (autoChainActive) source = 'auto_chain';
else if (autoAdvance) source = 'auto_advance';
output({
active: autoChainActive || autoAdvance,
source,
auto_chain_active: autoChainActive,
auto_advance: autoAdvance,
}, raw, undefined);
}
function gateEnabled(projectDir: string): boolean {
const value = readWorkflowConfig(projectDir).context_coverage_gate;
if (typeof value === 'boolean') return value;
if (typeof value === 'string') {
const lower = value.toLowerCase();
if (lower === 'false' || lower === 'true') return lower !== 'false';
}
return true;
}
function loadPlanContents(phaseDir: string): string[] {
if (!fs.existsSync(phaseDir)) return [];
try {
return fs.readdirSync(phaseDir)
.filter((entry) => /-PLAN\.md$/.test(entry))
.map((entry) => readIfExists(path.join(phaseDir, entry)));
} catch {
return [];
}
}
const DESIGNATED_HEADINGS_RE = /^#{1,6}\s+(?:must[_ ]haves?|truths?|tasks?|objective)\b/i;
const XML_DECISION_TAGS_RE = /<(?:objective|tasks?|action)(?:\s[^>]{0,1000})?>((?:(?!<(?:objective|tasks?|action)[\s>])[\s\S])*?)<\/(?:objective|tasks?|action)>/gi;
function stripCommentsAndFences(text: string): string {
// HTML-comment stripping stays caller-side (the seam does not strip HTML comments).
// Stop-at-next-open body (ReDoS-safe, #2128); an UNCLOSED `<!--` does not match,
// so downstream tags are preserved (unlike a `(?:-->|$)` fallback, which would
// wipe to EOF and fail-close the decision-coverage gate).
const htmlStripped = text.replace(/<!--(?:(?!<!--)[\s\S])*?-->/g, ' ');
// Fenced-code stripping: delegate to the canonical CommonMark-correct seam.
// replaces the prior independent regex copy (```` ``` ``` ```` + `~~~ ~~~`).
return stripFencedCode(htmlStripped).text;
}
function extractYamlBlock(frontmatter: string, key: string): string {
const match = frontmatter.match(new RegExp(`^${key}\\s*:(.*)$`, 'm'));
if (!match) return '';
const startIdx = (match.index || 0) + match[0].length;
const rest = frontmatter.slice(startIdx + 1).split(/\r?\n/);
const block = [match[1] || ''];
for (const line of rest) {
if (line === '' || /^\s/.test(line)) block.push(line);
else break;
}
return block.join('\n');
}
function extractXmlTagBodies(text: string): string {
const parts: string[] = [];
for (const match of text.matchAll(XML_DECISION_TAGS_RE)) {
if (match[1]) parts.push(match[1]);
}
return parts.join('\n');
}
function extractPlanDesignatedSections(planContent: string | null | undefined): string {
if (!planContent) return '';
const cleaned = stripCommentsAndFences(planContent);
const fmMatch = cleaned.match(/^---\r?\n([\s\S]*?)\r?\n---\r?\n?([\s\S]*)$/);
const frontmatter = fmMatch ? fmMatch[1] : '';
const body = fmMatch ? fmMatch[2] : cleaned;
const parts: string[] = [];
for (const key of ['must_haves', 'truths', 'objective']) {
const block = extractYamlBlock(frontmatter, key);
if (block) parts.push(block);
}
// Replace hand-rolled split(/\r?\n/) + heading walk with the seam's collectSections.
// stopPredicate fires on EVERY heading (collectSections needs to start a section at
// each heading), then we filter to designated ones — same semantics as the prior
// inDesignated flag: emit the heading line + body only when DESIGNATED_HEADINGS_RE matches.
const sections = collectSections(body, () => true);
const bodyParts: string[] = [];
for (const section of sections) {
const headingLine = '#'.repeat(section.heading.level) + ' ' + section.heading.text;
if (DESIGNATED_HEADINGS_RE.test(headingLine)) {
bodyParts.push(headingLine);
if (section.body) bodyParts.push(section.body);
}
}
parts.push(bodyParts.join('\n'));
parts.push(extractXmlTagBodies(cleaned));
return parts.join('\n\n');
}
interface UncoveredItem {
id: string;
text: string;
category: string;
}
function buildPlanMessage(uncovered: UncoveredItem[]): string {
if (uncovered.length === 0) return 'All trackable CONTEXT.md decisions are covered by plans.';
return [
'## Decision Coverage Gap',
'',
`${uncovered.length} CONTEXT.md decision(s) are not covered by any plan:`,
'',
...uncovered.map((item) => `- **${item.id}** (${item.category || 'uncategorized'}): ${item.text}`),
'',
'Resolve by citing `D-NN:` in a relevant plan\'s `must_haves`/`truths` (or body),',
'OR move the decision to `### Claude\'s Discretion` / tag it `[informational]` if it should not be tracked.',
].join('\n');
}
function buildVerifyMessage(notHonored: UncoveredItem[]): string {
if (notHonored.length === 0) return 'All trackable CONTEXT.md decisions are honored by shipped artifacts.';
return [
'### Decision Coverage (warning)',
'',
`${notHonored.length} decision(s) not found in shipped artifacts:`,
'',
...notHonored.map((item) => `- **${item.id}** (${item.category || 'uncategorized'}): ${item.text}`),
'',
'This is a soft warning - verification status is unchanged.',
].join('\n');
}
function loadDecisionExtraction(contextPath: string): { trackable: Decision[]; outcome: 'parsed' | 'none-present' | 'could-not-parse' } {
const extraction = extractDecisions(readIfExists(contextPath));
return {
trackable: extraction.decisions.filter((d) => d.trackable),
outcome: extraction.outcome,
};
}
function cmdDecisionCoveragePlan(projectDir: string, args: string[], raw: boolean): void {
const phaseDir = args[2] ? resolvePath(args[2], projectDir) : '';
const contextPath = args[3] ? resolvePath(args[3], projectDir) : '';
if (!gateEnabled(projectDir)) {
output({ passed: true, skipped: true, reason: 'workflow.context_coverage_gate is false', total: 0, covered: 0, uncovered: [], message: 'Decision coverage gate disabled by config.' }, raw, undefined);
return;
}
if (!contextPath || !fs.existsSync(contextPath)) {
output({ passed: true, skipped: true, reason: 'CONTEXT.md missing', total: 0, covered: 0, uncovered: [], message: 'No CONTEXT.md - nothing to check.' }, raw, undefined);
return;
}
const { trackable: decisions, outcome } = loadDecisionExtraction(contextPath);
// #1365 fail-loud gate: any could-not-parse outcome must NOT silently pass —
// even when some decisions were extracted (e.g. D-01 valid but D-02 malformed).
// A parse-miss on ANY bullet means the gate cannot certify full coverage.
// Fire independent of decisions.length so a partial-parse still blocks.
if (outcome === 'could-not-parse') {
const partialParse = decisions.length > 0;
output({
passed: false,
skipped: false,
reason: 'could-not-parse',
total: decisions.length,
covered: 0,
uncovered: [],
message: partialParse
? 'Decision coverage gate: decisions could not be fully parsed — one or more ' +
'`- **D-NN ...**` bullets appear malformed (missing `:` or ` — ` separator). ' +
'Fix the bullet format so all D-NN decisions can be read before re-running the gate.'
: 'Decision coverage gate: could not parse decisions — possible format mismatch. ' +
'The CONTEXT.md appears to be decision-shaped (has a <decisions> block, a decisions heading, ' +
'or D- tokens) but no D-NN bullets could be extracted. Check the formatting of the decisions ' +
'block and ensure bullets follow the `- **D-NN:** text` or `- **D-NN — title** body` form.',
}, raw, undefined);
return;
}
if (decisions.length === 0) {
output({ passed: true, skipped: true, reason: 'no trackable decisions', total: 0, covered: 0, uncovered: [], message: 'No trackable decisions in CONTEXT.md.' }, raw, undefined);
return;
}
const sections = loadPlanContents(phaseDir).map(extractPlanDesignatedSections);
const uncovered: UncoveredItem[] = [];
let covered = 0;
for (const decision of decisions) {
if (sections.some((section) => decisionMentioned(section, decision))) covered++;
else uncovered.push({ id: decision.id, text: decision.text, category: decision.category });
}
output({
passed: uncovered.length === 0,
skipped: false,
total: decisions.length,
covered,
uncovered,
message: buildPlanMessage(uncovered),
}, raw, undefined);
}
function recentCommitMessages(projectDir: string): string {
try {
return execFileSync('git', ['log', '-n', '200', '--pretty=%s%n%b'], {
cwd: projectDir,
encoding: 'utf-8',
maxBuffer: 4 * 1024 * 1024,
windowsHide: true,
});
} catch {
return '';
}
}
function isInsideRoot(candidatePath: string, rootDir: string): boolean {
const root = path.resolve(rootDir);
const target = path.resolve(root, candidatePath);
return target === root || target.startsWith(`${root}${path.sep}`);
}
function readModifiedFilesContent(projectDir: string, summaries: string[]): string {
const out: string[] = [];
let total = 0;
for (const summary of summaries) {
if (!summary) continue;
for (const blockMatch of summary.matchAll(/files_modified:\s*\n((?:[ \t]*-\s+.+\n?)+)/g)) {
const files = [...(blockMatch[1] || '').matchAll(/-\s+(.+)/g)]
.map((match) => match[1].trim().replace(/^["']|["']$/g, ''));
for (const file of files) {
if (total >= 50) break;
if (!file || !isInsideRoot(file, projectDir)) continue;
const raw = readIfExists(resolvePath(file, projectDir));
out.push(raw.length > 256 * 1024 ? raw.slice(0, 256 * 1024) : raw);
total++;
}
if (total >= 50) break;
}
if (total >= 50) break;
}
return out.join('\n\n');
}
function cmdDecisionCoverageVerify(projectDir: string, args: string[], raw: boolean): void {
const phaseDir = args[2] ? resolvePath(args[2], projectDir) : '';
const contextPath = args[3] ? resolvePath(args[3], projectDir) : '';
if (!gateEnabled(projectDir)) {
output({ skipped: true, blocking: false, reason: 'workflow.context_coverage_gate is false', total: 0, honored: 0, not_honored: [], message: 'Decision coverage gate disabled by config.' }, raw, undefined);
return;
}
if (!contextPath || !fs.existsSync(contextPath)) {
output({ skipped: true, blocking: false, reason: 'CONTEXT.md missing', total: 0, honored: 0, not_honored: [], message: 'No CONTEXT.md - nothing to check.' }, raw, undefined);
return;
}
const { trackable: decisions, outcome: decisionOutcome } = loadDecisionExtraction(contextPath);
// Mirror could-not-parse surface for verify (non-blocking advisory WARN).
// Fire independent of decisions.length — a parse-miss on any bullet must surface,
// even when some decisions were partially extracted (#1365 fix-parity with plan gate).
if (decisionOutcome === 'could-not-parse') {
const partialParse = decisions.length > 0;
output({
skipped: false,
blocking: false,
reason: 'could-not-parse',
total: decisions.length,
honored: 0,
not_honored: [],
message: partialParse
? 'Decision coverage verify (warning): decisions could not be fully parsed — one or more ' +
'`- **D-NN ...**` bullets appear malformed. Fix the bullet format in the CONTEXT.md decisions block.'
: 'Decision coverage verify (warning): could not parse decisions — possible format mismatch. ' +
'Check the formatting of the CONTEXT.md decisions block.',
}, raw, undefined);
return;
}
if (decisions.length === 0) {
output({ skipped: true, blocking: false, reason: 'no trackable decisions', total: 0, honored: 0, not_honored: [], message: 'No trackable decisions in CONTEXT.md.' }, raw, undefined);
return;
}
const planContents = loadPlanContents(phaseDir);
const summaryParts = fs.existsSync(phaseDir)
? fs.readdirSync(phaseDir).filter((entry) => /-SUMMARY\.md$/.test(entry)).map((entry) => readIfExists(path.join(phaseDir, entry)))
: [];
const haystack = [
planContents.join('\n\n'),
summaryParts.join('\n\n'),
readModifiedFilesContent(projectDir, summaryParts),
recentCommitMessages(projectDir),
].join('\n\n');
const notHonored: UncoveredItem[] = [];
let honored = 0;
for (const decision of decisions) {
if (decisionMentioned(haystack, decision)) honored++;
else notHonored.push({ id: decision.id, text: decision.text, category: decision.category });
}
output({
skipped: false,
blocking: false,
total: decisions.length,
honored,
not_honored: notHonored,
message: buildVerifyMessage(notHonored),
}, raw, undefined);
}
// ─── ui-plan-gate ─────────────────────────────────────────────────────────────
/**
* ui-plan-gate: given a phase number, checks whether the phase has frontend
* indicators and whether a *-UI-SPEC.md already exists in the phase directory.
*
* Returns JSON: { frontend: boolean, hasUiSpec: boolean, block: boolean }
* block = frontend && !hasUiSpec (gate fires when UI work is detected but no spec exists)
*
* Invocable as: gsd_run check ui-plan-gate <phase>
*
* Uses checkUiPresence from ui-safety-gate.cjs — does NOT reimplement frontend detection.
* Uses getRoadmapPhaseWithFallback + findPhaseInternal from leaf modules for phase data.
*/
function findUiSpecInDir(phaseDir: string): string {
if (!phaseDir || !fs.existsSync(phaseDir)) return '';
try {
const files = fs.readdirSync(phaseDir);
const found = files.find((f) => /-UI-SPEC\.md$/.test(f));
return found ? path.join(phaseDir, found) : '';
} catch {
return '';
}
}
/**
* Pure logic for ui-plan-gate — exposed for direct behavioral testing.
*
* Given a projectDir and phase number:
* (a) Reads the phase section from ROADMAP.md via getRoadmapPhaseWithFallback —
* same two-pass lookup (current milestone → full roadmap) as `roadmap.get-phase`
* (cmdRoadmapGetPhase). Cross-milestone / older frontend phases resolve correctly.
* If ROADMAP.md is missing, phaseSection is '' (ROADMAP.md not present = project
* has no roadmap = cannot be frontend). If the phase truly can't be found after
* both passes, phaseSection is '' and phaseLookupFailed is set so callers can
* surface the miss — we do NOT silently degrade to frontend:false if the roadmap
* exists but the phase header is absent.
* (b) Runs checkUiPresence (frontend detection) — no reimplementation.
* (c) Resolves the phase directory via findPhaseInternal (phase-locator.cjs); checks for *-UI-SPEC.md.
*
* Returns: { frontend, hasUiSpec, block, uiSpecPath, phaseLookupFailed }
* block = frontend && !hasUiSpec
* phaseLookupFailed = ROADMAP.md present but phase header not found (surfaced for
* onError:halt gates so a missing phase doesn't silently bypass)
*/
function computeUiPlanGate(projectDir: string, phase: string): {
frontend: boolean;
hasUiSpec: boolean;
block: boolean;
uiSpecPath: string | null;
phaseLookupFailed?: boolean;
} {
// (a) Read the phase section text using the same two-pass lookup as roadmap.get-phase.
// getRoadmapPhaseWithFallback: current-milestone first, then stripShippedMilestones
// fallback — mirrors cmdRoadmapGetPhase exactly.
let phaseSection = '';
let phaseLookupFailed: boolean | undefined;
try {
const section = getRoadmapPhaseWithFallback(projectDir, phase);
if (section === null) {
// Distinguish: ROADMAP.md missing (no-roadmap project) vs phase not found in ROADMAP.
// planningDir(cwd) resolves the .planning/ root for workstream-aware paths.
const planDir: string = planningDir(projectDir);
const roadmapPath = path.join(planDir, 'ROADMAP.md');
if (fs.existsSync(roadmapPath)) {
// ROADMAP.md exists but phase was not found → surface the miss
phaseLookupFailed = true;
}
// phaseSection stays ''
} else {
phaseSection = section;
}
} catch { /* roadmap read failure → treat as empty (non-frontend) */ }
// (b) Run checkUiPresence (frontend detection) — reuse existing helper; no reimplementation
const presenceResult = checkUiPresence(phaseSection);
const frontend = presenceResult.hasUI;
// (c) Resolve phase directory via findPhaseInternal and check for *-UI-SPEC.md
let phaseDir = '';
try {
const result = findPhaseInternal(projectDir, phase);
if (result && typeof result === 'object') {
// findPhaseInternal returns { directory: '<relative-posix-path>', ... }
// directory is relative to cwd — resolve it to absolute.
const relDir = typeof result['directory'] === 'string' ? result['directory'] : '';
if (relDir) {
phaseDir = path.resolve(projectDir, relDir);
}
} else if (typeof result === 'string') {
phaseDir = result;
}
} catch { /* phase dir lookup failure → hasUiSpec=false */ }
const uiSpecPath = findUiSpecInDir(phaseDir);
const hasUiSpec = uiSpecPath !== '';
// block = frontend phase with no UI-SPEC
const block = frontend && !hasUiSpec;
const result: { frontend: boolean; hasUiSpec: boolean; block: boolean; uiSpecPath: string | null; phaseLookupFailed?: boolean } = {
frontend, hasUiSpec, block, uiSpecPath: hasUiSpec ? uiSpecPath : null,
};
if (phaseLookupFailed) result.phaseLookupFailed = true;
return result;
}
function cmdUiPlanGate(projectDir: string, args: string[], raw: boolean): void {
// args[0] = 'check', args[1] = 'ui-plan-gate', args[2] = phase
const phase = args[2] || '';
if (!phase) {
error('ui-plan-gate requires a phase argument: check ui-plan-gate <phase>', ERROR_REASON.SDK_MISSING_ARG);
return;
}
output(computeUiPlanGate(projectDir, phase), raw, undefined);
}
// ─── ui-safety-gate ───────────────────────────────────────────────────────────
/**
* ui-safety-gate: post-wave check that verifies UI-changed files conform to
* the active UI-SPEC for the phase. Called after each wave by execute:wave:post.
*
* Returns JSON: { frontend: boolean, hasUiFiles: boolean, hasUiSpec: boolean, block: boolean, message?: string }
* block = frontend && hasUiFiles && !hasUiSpec
*
* Args: check ui-safety-gate <phase>
* Invocable as: gsd_run check ui-safety-gate <phase>
* or gsd_run check ui.safety-gate <phase> (dots normalized to hyphens)
*
* Uses checkUiPresence from ui-safety-gate.cjs — does NOT reimplement frontend detection.
* Checks whether any files changed in recent git history match frontend file patterns.
* Also checks whether a *-UI-SPEC.md exists in the phase directory (same as ui-plan-gate).
*
* Limitation: uses git diff HEAD~1..HEAD which covers only the last commit; in a
* multi-plan wave the wave-start commit would be more accurate but is not yet stored
* in the wave manifest. This is tracked as a known limitation.
*/
const UI_FILE_EXTENSIONS_RE = /\.(tsx|jsx|css|scss|sass|less|vue|svelte|html)$/i;
const UI_PATH_PATTERNS_RE = /\/(components|pages|views|screens|layouts|ui|frontend)\//i;
/**
* Pure logic for ui-safety-gate — exposed for direct behavioral testing.
*
* Given a projectDir and phase number:
* (a) Reads the phase section from ROADMAP.md via getRoadmapPhaseWithFallback —
* same lookup as computeUiPlanGate — to determine if this is a frontend phase.
* (b) Runs checkUiPresence (frontend detection) — no reimplementation.
* (c) Checks git diff HEAD~1..HEAD for UI file changes in the current worktree.
* (d) Resolves the phase directory via findPhaseInternal (phase-locator.cjs); checks for *-UI-SPEC.md.
*
* Returns: { frontend, hasUiFiles, hasUiSpec, block, message?, phaseLookupFailed? }
* block = frontend && hasUiFiles && !hasUiSpec
* phaseLookupFailed = ROADMAP.md present but phase header not found
*/
function computeUiSafetyGate(projectDir: string, phase: string): {
frontend: boolean;
hasUiFiles: boolean;
hasUiSpec: boolean;
block: boolean;
message?: string;
phaseLookupFailed?: boolean;
} {
// (a) Read the phase section text (same two-pass lookup as computeUiPlanGate)
let phaseSection = '';
let phaseLookupFailed: boolean | undefined;
try {
const section = getRoadmapPhaseWithFallback(projectDir, phase);
if (section === null) {
const planDir: string = planningDir(projectDir);
const roadmapPath = path.join(planDir, 'ROADMAP.md');
if (fs.existsSync(roadmapPath)) {
phaseLookupFailed = true;
}
} else {
phaseSection = section;
}
} catch { /* roadmap read failure → treat as empty (non-frontend) */ }
// (b) Run checkUiPresence (frontend detection) — reuse existing helper; no reimplementation
const presenceResult = checkUiPresence(phaseSection);
const frontend = presenceResult.hasUI;
// (c) Check whether any UI files were changed in recent git commits
// Uses git diff HEAD~1..HEAD to detect frontend file changes since last commit.
// Known limitation: multi-plan waves may need the wave-start commit for full coverage.
let hasUiFiles = false;
try {
const changed = execFileSync('git', ['diff', '--name-only', 'HEAD~1', 'HEAD'], {
cwd: projectDir,
encoding: 'utf-8',
maxBuffer: 2 * 1024 * 1024,
windowsHide: true,
});
hasUiFiles = changed.split('\n').some((f) =>
f.trim() && (UI_FILE_EXTENSIONS_RE.test(f) || UI_PATH_PATTERNS_RE.test(f)),
);
} catch { /* git unavailable or no prior commit — treat as no UI files changed */ }
// (d) Resolve phase directory and check for *-UI-SPEC.md (same as computeUiPlanGate)
let phaseDir = '';
try {
const result = findPhaseInternal(projectDir, phase);
if (result && typeof result === 'object') {
const relDir = typeof result['directory'] === 'string' ? result['directory'] : '';
if (relDir) {
phaseDir = path.resolve(projectDir, relDir);
}
} else if (typeof result === 'string') {
phaseDir = result;
}
} catch { /* phase dir lookup failure → hasUiSpec=false */ }
const uiSpecPath = findUiSpecInDir(phaseDir);
const hasUiSpec = uiSpecPath !== '';
// block only when: this is a frontend phase AND UI files were changed AND no UI-SPEC exists
const block = frontend && hasUiFiles && !hasUiSpec;
const result: {
frontend: boolean;
hasUiFiles: boolean;
hasUiSpec: boolean;
block: boolean;
message?: string;
phaseLookupFailed?: boolean;
} = { frontend, hasUiFiles, hasUiSpec, block };
if (block) {
result.message = `UI files changed in this wave but no UI-SPEC.md exists for Phase ${phase}. ` +
`Run /gsd:ui-phase ${phase} to generate the design contract before continuing.`;
}
if (phaseLookupFailed) result.phaseLookupFailed = true;
return result;
}
function cmdUiSafetyGate(projectDir: string, args: string[], raw: boolean): void {
// args[0] = 'check', args[1] = 'ui-safety-gate', args[2] = phase
const phase = args[2] || '';
if (!phase) {
error('ui-safety-gate requires a phase argument: check ui-safety-gate <phase>', ERROR_REASON.SDK_MISSING_ARG);
return;
}
output(computeUiSafetyGate(projectDir, phase), raw, undefined);
}
// ─── tdd-review-checkpoint ────────────────────────────────────────────────────
/**
* tdd-review-checkpoint: end-of-phase advisory check that scans type:tdd plans
* for RED/GREEN/REFACTOR gate-sequence compliance and surfaces a review table.
*
* Logic from gsd-core/references/tdd.md <end_of_phase_review> and
* execute-phase.md <step name="tdd_review_checkpoint"> (now removed).
*
* Returns JSON:
* { passed: true, tddPlans: N, violations: N, table: string, rows: PlanRow[] }
* where passed is always true (advisory gate — never blocks).
*
* Args: check tdd.review-checkpoint <phase>
* Phase can be a number or phase-dir path; if not resolvable the check
* returns passed:true with tddPlans:0 (no plans to review).
*/
interface TddPlanRow {
planId: string;
red: boolean;
green: boolean;
refactor: boolean;
status: 'Pass' | 'FAIL';
missing: string[];
}
function cmdTddReviewCheckpoint(projectDir: string, args: string[], raw: boolean): void {
// args[0] = 'check', args[1] = 'tdd-review-checkpoint' (normalized), args[2] = phase
const phase = args[2] || '';
if (!phase) {
error('tdd.review-checkpoint requires a phase argument: check tdd.review-checkpoint <phase>', ERROR_REASON.SDK_MISSING_ARG);
return;
}
// Resolve phase directory
let phaseDir = '';
try {
const result = findPhaseInternal(projectDir, phase);
if (result && typeof result === 'object') {
const relDir = typeof result['directory'] === 'string' ? result['directory'] : '';
if (relDir) phaseDir = path.resolve(projectDir, relDir);
} else if (typeof result === 'string') {
phaseDir = result;
}
} catch { /* phase dir lookup failure */ }
// Find all PLAN.md files with type: tdd in frontmatter
const tddPlanFiles: string[] = [];
if (phaseDir) {
try {
const files = fs.readdirSync(phaseDir).filter(f => f.endsWith('-PLAN.md'));
for (const file of files) {
const planPath = path.join(phaseDir, file);
const content = readIfExists(planPath);
// Check frontmatter for type: tdd
const frontmatterMatch = content.match(/^---\n([\s\S]*?)\n---/);
if (frontmatterMatch) {
const fm = frontmatterMatch[1];
if (/^type:\s*tdd\s*$/m.test(fm)) {
tddPlanFiles.push(planPath);
}
}
}
} catch { /* directory read failure */ }
}
if (tddPlanFiles.length === 0) {
const result = {
// Uniform gate contract: block = violations > 0 (advisory; never truly blocks).
block: false,
passed: true,
tddPlans: 0,
violations: 0,
table: '',
rows: [] as TddPlanRow[],
message: `No type:tdd plans found in phase ${phase}. TDD review skipped.`,
};
// Pass undefined as rawValue so --raw emits JSON (not plain text).
// The human-readable report is carried in `result.message` for the
// dispatch's advisory branch to surface.
output(result, raw, undefined);
return;
}
// For each TDD plan, extract the plan ID (padded plan number) and check git log
const rows: TddPlanRow[] = [];
for (const planPath of tddPlanFiles) {
// Extract plan ID from filename (e.g. "01-02-PLAN.md" → "01-02", or "03-PLAN.md" → "03")
const basename = path.basename(planPath, '-PLAN.md');
// planId for commit grep: phase-plan format, e.g. "01-02"
const planId = basename;
// Check for RED gate commit: test({planId}):
let red = false;
let green = false;
let refactor = false;
try {
const redCommit = execFileSync(
'git', ['log', '--oneline', `--grep=^test(${planId}):`, '--', '.'],
{ cwd: projectDir, encoding: 'utf-8', maxBuffer: 1024 * 1024, windowsHide: true },
);
red = redCommit.trim().length > 0;
} catch { /* git unavailable or no match */ }
try {
const greenCommit = execFileSync(
'git', ['log', '--oneline', `--grep=^feat(${planId}):`, '--', '.'],
{ cwd: projectDir, encoding: 'utf-8', maxBuffer: 1024 * 1024, windowsHide: true },
);
green = greenCommit.trim().length > 0;
} catch { /* git unavailable or no match */ }
try {
const refactorCommit = execFileSync(
'git', ['log', '--oneline', `--grep=^refactor(${planId}):`, '--', '.'],
{ cwd: projectDir, encoding: 'utf-8', maxBuffer: 1024 * 1024, windowsHide: true },
);
refactor = refactorCommit.trim().length > 0;
} catch { /* git unavailable or no match */ }
const missing: string[] = [];
if (!red) missing.push('RED');
if (!green) missing.push('GREEN');
const status: 'Pass' | 'FAIL' = missing.length === 0 ? 'Pass' : 'FAIL';
rows.push({ planId, red, green, refactor, status, missing });
}
const violations = rows.filter(r => r.status === 'FAIL').length;
// Build review table
const sep = '━'.repeat(53);
const tableHeader = '| Plan | RED | GREEN | REFACTOR | Status |';
const tableDivider = '|------|-----|-------|----------|--------|';
const tableRows = rows.map(r =>
`| ${r.planId.padEnd(4)} | ${r.red ? ' ✓ ' : ' ✗ '} | ${r.green ? ' ✓ ' : ' ✗ '} | ${r.refactor ? ' ✓ ' : ' — '} | ${r.status.padEnd(6)} |`,
);
let table = [
sep,
` TDD REVIEW — Phase ${phase}`,
sep,
'',
`TDD Plans: ${tddPlanFiles.length} | Gate violations: ${violations}`,
'',
tableHeader,
tableDivider,
...tableRows,
].join('\n');
if (violations > 0) {
table += '\n\n⚠ Gate violations are advisory — review before advancing.';
for (const r of rows.filter(row => row.status === 'FAIL')) {
table += `\n Plan ${r.planId} missing: ${r.missing.join(', ')} gate commit(s).`;
table += `\n Expected commit pattern: test(${r.planId}): ... → feat(${r.planId}): ...`;
}
}
const result = {
// Uniform gate contract: block = violations > 0.
// This gate is advisory (blocking: false in capability.json) so block:true
// only surfaces as a warning, never halts. Kept here so the host-loop
// dispatch can read a single consistent `block` field.
block: violations > 0,
passed: true,
tddPlans: tddPlanFiles.length,
violations,
table,
rows,
// Human-readable report in `message` so the dispatch's advisory branch
// can surface it. --raw emits JSON (rawValue=undefined), not plain text.
message: table,
};
// Pass undefined as rawValue so --raw emits JSON (not the raw table text).
// The review table is carried in `result.message` and `result.table` so
// the host-loop dispatch's advisory branch can surface it.
output(result, raw, undefined);
}
// ─── gap-analysis-plan-post ───────────────────────────────────────────────────
/**
* gap-analysis-plan-post: non-blocking advisory check that runs the post-planning
* gap analysis after all PLAN.md files are generated for a phase.
*
* Cross-references every REQ-ID and D-ID from REQUIREMENTS.md and CONTEXT.md
* against the concatenated text of all *-PLAN.md files, emitting a coverage table.
*
* This gate is always advisory (passed: true) — it never blocks phase advancement.
*
* Args: check gap-analysis.plan-post <phase-dir> [phase-req-ids]
* Invocable as: gsd_run check gap-analysis.plan-post <phase-dir> [phase-req-ids]
*/
function cmdGapAnalysisPlanPost(projectDir: string, args: string[], raw: boolean): void {
// args[0] = 'check', args[1] = 'gap-analysis-plan-post' (normalized), args[2] = phaseDir, args[3] = phaseReqIds
const phaseDir = args[2] || '';
if (!phaseDir) {
error('gap-analysis.plan-post requires a phase-dir argument: check gap-analysis.plan-post <phase-dir> [phase-req-ids]', ERROR_REASON.SDK_MISSING_ARG);
return;
}
const phaseReqIds = args[3] ?? undefined;
const result = runGapAnalysis(projectDir, phaseDir, { phaseReqIds });
// Uniform gate contract: block = false (gap-analysis is always advisory, never blocks).
// `message` carries the human-readable gap analysis report so the dispatch's
// advisory branch can surface it. --raw emits JSON (rawValue=undefined), not
// plain markdown text.
output(
{
block: false,
passed: true,
enabled: result.enabled,
table: result.table,
summary: result.summary,
counts: result.counts,
// Human-readable report in `message` for the host-loop advisory branch.
message: result.table || result.summary || '',
},
raw,
undefined,
);
}
interface RouteCheckCommandOptions {
args: string[];
cwd: string;
raw: boolean;
}
// ─── predicate (generic gate-predicate evaluator, #2008) ──────────────────────
/**
* Production subprocess binding for the gate-predicate evaluator. Wraps the
* bounded `execTool` seam (shell-command-projection) as a `runBoundedShell`
* the pure evaluator consumes. `sh -c` runs the interpolated command; the
* subprocess inherits the process env and is killed (SIGTERM) on timeout.
*
* `timedOut` is derived from the kill signal: spawnSync sets `signal: 'SIGTERM'`
* when the `timeout` fires, distinct from a normal non-zero exit code. A command
* that self-terminates with SIGTERM is indistinguishable at this seam and is
* reported as a timeout — either way the gate blocks (non-zero), so the outcome
* is fail-closed and correct. See ADR-2008.
*/
function buildPredicateDeps() {
return {
runBoundedShell(opts: { command: string; cwd: string; timeoutMs: number }): {
exitCode: number | null;
stdout: string;
stderr: string;
signal: NodeJS.Signals | null;
timedOut: boolean;
} {
const r = execTool('sh', ['-c', opts.command], { cwd: opts.cwd, timeout: opts.timeoutMs });
return {
exitCode: r.exitCode,
stdout: r.stdout,
stderr: r.stderr,
signal: r.signal,
timedOut: r.signal === 'SIGTERM',
};
},
};
}
/** Parse `--flag value` pairs from an args array into a map (last write wins). */
function parsePredicateFlags(args: string[]): Record<string, string> {
const out: Record<string, string> = {};
for (let i = 0; i < args.length; i++) {
const a = args[i];
if (typeof a !== 'string') continue;
if (!a.startsWith('--')) continue;
const key = a.slice(2);
const next = args[i + 1];
if (key.length > 0 && typeof next === 'string' && !next.startsWith('--')) {
out[key] = next;
i++;
}
}
return out;
}
/**
* `check predicate` — generic evaluator for capability gate `check.predicate`
* blocks (#2008). The workflow gate-dispatch invokes this for any gate whose
* `check` carries a `predicate` (instead of a `query`); the predicate object is
* passed as `--predicate '<json>'`. Emits the standard `{ block, message,
* details? }` gate contract on success. A malformed predicate / unknown kind
* THROWS inside the evaluator and is mapped here to `error()` (non-zero exit),
* which the workflow's two-step gate contract treats as a step-1 command failure
* routed per the gate's `onError`.
*
* Invocation:
* gsd_run check predicate --predicate '<json>' \
* [--phase-dir <dir>] [--phase-number <n>] [--phase-req-ids <ids>] --raw
*
* The subprocess runs at the runtime project root (the `cwd` passed to this
* router), inheriting the process env. Interpolation placeholders
* ${PHASE_NUMBER}/${PHASE_DIR}/${PHASE_REQ_IDS} are substituted from the flags.
*/
function cmdCheckPredicate(projectDir: string, args: string[], raw: boolean): void {
const flags = parsePredicateFlags(args);
const predicateJson = flags['predicate'];
if (!predicateJson) {
error('predicate requires --predicate <json> (the gate hook check.predicate object)', ERROR_REASON.SDK_MISSING_ARG);
return;
}
let predicate: unknown;
try {
predicate = JSON.parse(predicateJson);
} catch {
error('predicate --predicate value must be valid JSON', ERROR_REASON.USAGE);
return;
}
const ctx = {
cwd: projectDir,
phaseNumber: flags['phase-number'],
phaseDir: flags['phase-dir'],
phaseReqIds: flags['phase-req-ids'],
};
let result;
try {
result = evaluatePredicate(predicate, ctx, buildPredicateDeps());
} catch (e) {
error(`gate predicate evaluation failed: ${(e as Error).message}`, ERROR_REASON.USAGE);
return;
}
output(result, raw, undefined);
}
// ─── api-coverage-verify-pre ──────────────────────────────────────────────────
/**
* api-coverage.verify-pre: BLOCKING seal-time gate for the ai-integration
* capability (#1562). Enforces "Full API Coverage by Default — Opt Out, Never
* Opt In." A phase that integrates an external API/SDK/service may not seal
* until a COVERAGE.md matrix enumerates the surface and every non-integrated
* capability is an explicit, reasoned opt-out.
*
* Contract (two touch points composed into one check):
* 1. If COVERAGE.md exists in the phase dir → validate it (acceptance #2).
* Block on any validation error (empty matrix, OPT-OUT without reason,
* duplicate/empty capability).
* 2. If COVERAGE.md is absent → run detectApiIntegration over the phase scope
* (PLAN.md body, then ROADMAP phase section as fallback). If a strong
* external-API-integration signal is detected → BLOCK ("integration
* detected without coverage matrix"). If no signal → PASS (treat as a
* non-API phase; acceptance #4 — low false positives).
*
* The detector is the FALLBACK for the "nobody decided / forgot the matrix"
* case; the primary path is the plan:pre contribution prompting COVERAGE.md.
*
* Args: check api-coverage.verify-pre <phase-dir>
* Emits the uniform gate contract: { block, passed, message, ...details }.
*/
function cmdApiCoverageVerifyPre(projectDir: string, args: string[], raw: boolean): void {
const phaseArg = typeof args[2] === 'string' ? args[2] : '';
if (!phaseArg) {
error(
'api-coverage.verify-pre requires a phase argument: check api-coverage.verify-pre <phase-dir-or-token>',
ERROR_REASON.SDK_MISSING_ARG,
);
return;
}
const pDir = planningDir(projectDir);
const phasesRoot = path.join(pDir, 'phases');
// SECURITY (path traversal): the phase argument is taken ONLY as a phase
// token — its basename — and resolved by findPhaseInternal strictly under
// .planning/phases/ (or a milestone archive). The raw arg is never used as a
// path, so `..`, absolute paths, and arbitrary directories cannot reach a
// file read. Mirrors cmdVerifySchemaDrift's token-match approach.
let token = posixNormalize(phaseArg).split('/').filter(Boolean).pop() || '';
// A token like ".." or "." carries no phase identity → unresolvable.
if (token === '.' || token === '..') token = '';
// Not a GSD project (no phases tree at all) → fail-open: nothing to gate.
if (!fs.existsSync(phasesRoot)) {
output(
{
block: false,
passed: true,
coverage_present: false,
detected: false,
message: 'api-coverage: no .planning/phases directory; gate skipped (not a GSD project layout)',
},
raw,
undefined,
);
return;
}
// Resolve the phase dir under the contained phases root.
let resolvedDir: string | null = null;
let phaseNumber = '';
if (token) {
const found = findPhaseInternal(projectDir, token);
if (found && found.directory) {
resolvedDir = found.directory;
phaseNumber = found.phase_number || '';
}
}
if (!resolvedDir) {
// The phases tree EXISTS but THIS phase could not be resolved. For a
// BLOCKING gate, fail-closed: a missing phase dir must not silently bypass
// the coverage requirement. (Distinguished from "no .planning at all"
// above, which is a genuine non-GSD-project → pass.)
output(
{
block: true,
passed: false,
coverage_present: false,
detected: false,
phase_lookup_failed: true,
message:
`api-coverage: could not resolve phase "${phaseArg}" under .planning/phases/. ` +
'Resolve the phase directory (or produce COVERAGE.md) before sealing.',
},
raw,
undefined,
);
return;
}
// Defense-in-depth: the resolved dir must be inside the phases root (or a
// milestone archive under .planning/milestones).
const milestonesRoot = path.join(pDir, 'milestones');
if (!isInsideRoot(resolvedDir, phasesRoot) && !isInsideRoot(resolvedDir, milestonesRoot)) {
output(
{
block: true,
passed: false,
coverage_present: false,
detected: false,
message: 'api-coverage: resolved phase dir escapes .planning/ — refusing to evaluate',
},
raw,
undefined,
);
return;
}
// (1) locate COVERAGE.md — prefer the exact name, then a single *-COVERAGE.md.
let coverageFile = '';
let suffixed: string[] = [];
try {
const entries = fs.readdirSync(resolvedDir, { withFileTypes: true });
const files = entries.filter((e) => e.isFile()).map((e) => e.name);
const exact = files.find((f) => /^COVERAGE\.md$/i.test(f));
if (exact) {
coverageFile = exact;
} else {
suffixed = files.filter((f) => /-COVERAGE\.md$/i.test(f)).sort();
if (suffixed.length === 1) coverageFile = suffixed[0];
}
} catch {
// readdir failure → treat as no matrix readable; fall through to detection.
}
if (coverageFile) {
let matrixText: string;
try {
matrixText = fs.readFileSync(path.join(resolvedDir, coverageFile), 'utf8');
} catch {
// COVERAGE.md exists but is unreadable (EACCES/EIO/encoding). Fail-closed
// with a useful message rather than a raw throw.
output(
{
block: true,
passed: false,
coverage_present: true,
message: `api-coverage: COVERAGE.md exists but is unreadable — fix file permissions/encoding before sealing`,
},
raw,
undefined,
);
return;
}
const v = validateCoverageMatrix(matrixText);
if (v.valid) {
if (v.none_declared) {
// The declaration is the human override for the detector — it PASSES
// even when detection fires (that is acceptance #5's point: the
// detector is fallible and the declaration is the reasoned overrule).
// But a contradiction must be VISIBLE, not silent: re-run detection
// over the phase scope and surface any signals it still finds
// (#2365 review S-1).
const declScope = readPhaseScope(projectDir, resolvedDir, phaseNumber);
const declDetection = detectApiIntegration(declScope.text);
const declSignals = declDetection.signals.map((s) => ({ verb: s.verb, noun: s.noun }));
// The declaration legitimately wins even over a read error (it is the
// human overrule), but if scope was incomplete we say so — the contract
// is that contradictions stay visible, not silent (#2365 review).
const baseMsg = declDetection.detected
? `api-coverage: COVERAGE.md declares no external API integration, overriding ${declSignals.length} detected signal(s) — confirm the declaration is accurate`
: 'api-coverage: COVERAGE.md declares no external API integration — matrix not required';
output(
{
block: false,
passed: true,
coverage_present: true,
matrix: coverageFile,
counts: v.counts,
none_declared: true,
detected: declDetection.detected,
...(declDetection.detected ? { signals: declSignals } : {}),
...(declScope.readError ? { scope_read_error: declScope.readError } : {}),
message: declScope.readError
? `${baseMsg} (note: phase scope was incompletely read — ${declScope.readError})`
: baseMsg,
},
raw,
undefined,
);
return;
}
output(
{
block: false,
passed: true,
coverage_present: true,
matrix: coverageFile,
counts: v.counts,
message: `api-coverage: matrix present (${v.counts.surface} capabilities, ${v.counts.optout} opt-out)`,
},
raw,
undefined,
);
return;
}
// Fixed-template message (no raw cell content echoed into the LLM-facing
// message). The structured `errors` array is safe (row-indexed, no cell
// values) and travels as data for tooling that wants detail.
output(
{
block: true,
passed: false,
coverage_present: true,
matrix: coverageFile,
error_count: v.errors.length,
errors: v.errors,
message: `api-coverage: COVERAGE.md has ${v.errors.length} problem(s) — fix the matrix (every capability INTEGRATE or OPT-OUT with a reason) before sealing`,
},
raw,
undefined,
);
return;
}
if (suffixed.length > 1) {
output(
{
block: true,
passed: false,
coverage_present: false,
message: `api-coverage: multiple *-COVERAGE.md files found (${suffixed.length}) — consolidate into one COVERAGE.md before sealing`,
},
raw,
undefined,
);
return;
}
// (2) no matrix — detect whether this phase integrates an external API.
const scope = readPhaseScope(projectDir, resolvedDir, phaseNumber);
if (scope.readError) {
// Fail-closed: an unreadable plan could be the one describing the
// integration, so we cannot certify "no integration" — block and surface it.
output(
{
block: true,
passed: false,
coverage_present: false,
detected: false,
message:
`api-coverage: could not read the phase scope (${scope.readError}); ` +
'refusing to certify no external-API integration from incomplete scope. ' +
'Fix the unreadable plan file, or add a COVERAGE.md declaration.',
},
raw,
undefined,
);
return;
}
const detection = detectApiIntegration(scope.text);
if (detection.detected) {
// Surface only verb/noun (typed, bounded) — NOT raw prose snippets — so the
// gate output cannot relay injected PLAN.md instructions to the orchestrator.
const signals = detection.signals.map((s) => ({ verb: s.verb, noun: s.noun }));
output(
{
block: true,
passed: false,
coverage_present: false,
detected: true,
signals,
message:
'api-coverage: external-API integration detected without a coverage matrix. ' +
'Produce COVERAGE.md enumerating the API surface (every capability INTEGRATE or ' +
'OPT-OUT with a reason) before sealing. Full coverage is the default.',
},
raw,
undefined,
);
return;
}
output(
{
block: false,
passed: true,
coverage_present: false,
detected: false,
message: 'api-coverage: no external-API integration detected; coverage matrix not required',
},
raw,
undefined,
);
}
/**
* Read the phase-scope text used for API-integration detection. Uses the
* resolved plan files (PLAN.md bodies — the planner's own words about what the
* phase does) and, as a fallback, ONLY THIS PHASE'S ROADMAP section (not the
* whole roadmap, which would cross-contaminate sibling phases). Strips nothing
* here — detectApiIntegration strips fenced code itself.
*/
interface PhaseScopeRead {
text: string;
/** Non-null when a plan file EXISTED but could not be read. The gate must not
* conclude "no external API integration" from provably incomplete scope — an
* unreadable plan could be the one describing the integration (#2365 review:
* the blocking consumer silently passed partially-read scope). A missing plan
* directory is NOT a read error (a phase may legitimately have no plans yet). */
readError: string | null;
}
/** A filesystem error that is NOT "does not exist" — i.e. a real read failure
* (EACCES/EIO/…) the gate must not swallow. `ENOENT` is a legitimate "not
* there yet" and is treated as absence, not error. */
function isRealReadFailure(err: unknown): boolean {
const code = (err as NodeJS.ErrnoException | undefined)?.code;
return err != null && code !== 'ENOENT';
}
function readPhaseScope(projectDir: string, phaseDir: string, phaseNumber: string): PhaseScopeRead {
const chunks: string[] = [];
let readError: string | null = null;
try {
const entries = fs.readdirSync(phaseDir, { withFileTypes: true });
const plans = entries
.filter((e) => e.isFile() && /-PLAN\.md$/i.test(e.name))
.map((e) => e.name)
.sort();
for (const p of plans) {
try {
chunks.push(fs.readFileSync(path.join(phaseDir, p), 'utf8'));
} catch (err) {
// A plan file that exists but cannot be read — record it and keep
// reading the rest so the message names the first failure.
if (!readError) {
readError = `could not read ${p}: ${err instanceof Error ? err.message : String(err)}`;
}
}
}
} catch (err) {
// A MISSING phase directory is fine (no plans yet → fall through to the
// roadmap). A directory that exists but cannot be enumerated (EACCES/EIO)
// is a real read failure the gate must not silently pass (#2365 review).
if (isRealReadFailure(err)) {
return {
text: '',
readError: `could not read the phase directory: ${err instanceof Error ? err.message : String(err)}`,
};
}
}
if (readError) return { text: chunks.join('\n\n'), readError };
if (chunks.join('').trim().length > 0) return { text: chunks.join('\n\n'), readError: null };
// Fallback: ONLY this phase's ROADMAP section (not the whole file, which
// would pollute detection with sibling-phase prose). A MISSING roadmap/section
// is non-fatal; a roadmap that exists but cannot be read is a real failure.
if (phaseNumber) {
try {
const section = getRoadmapPhaseWithFallback(projectDir, phaseNumber);
if (section) return { text: section, readError: null };
} catch (err) {
if (isRealReadFailure(err)) {
return {
text: '',
readError: `could not read the roadmap fallback: ${err instanceof Error ? err.message : String(err)}`,
};
}
}
}
return { text: '', readError: null };
}
function routeCheckCommand({ args, cwd, raw }: RouteCheckCommandOptions): void {
// Normalize dots to hyphens in the subcommand so both forms are accepted.
// This makes `check.query = "ui.plan-gate"` (dotted form in capability.json gates)
// directly runnable as `gsd_run check ui.plan-gate` — the dot is normalized to
// `ui-plan-gate` before routing. The generic gate-dispatch in §5.6 reads
// `check.query` from the active gate hook and runs `gsd_run check ${hook.check.query}`,
// so the declared query must be dispatchable exactly as declared.
const rawSubcommand = args[1];
const subcommand = typeof rawSubcommand === 'string' ? rawSubcommand.replace(/\./g, '-') : rawSubcommand;
if (subcommand === 'auto-mode') {
cmdAutoMode(cwd, raw);
return;
}
if (subcommand === 'decision-coverage-plan') {
cmdDecisionCoveragePlan(cwd, args, raw);
return;
}
if (subcommand === 'decision-coverage-verify') {
cmdDecisionCoverageVerify(cwd, args, raw);
return;
}
if (subcommand === 'ui-plan-gate') {
cmdUiPlanGate(cwd, args, raw);
return;
}
if (subcommand === 'gap-analysis-plan-post') {
cmdGapAnalysisPlanPost(cwd, args, raw);
return;
}
if (subcommand === 'api-coverage-verify-pre') {
// ai-integration capability blocking gate at verify:pre (#1562). Dot-to-
// hyphen normalization means query "api-coverage.verify-pre" routes here.
cmdApiCoverageVerifyPre(cwd, args, raw);
return;
}
if (subcommand === 'tdd-review-checkpoint') {
cmdTddReviewCheckpoint(cwd, args, raw);
return;
}
if (subcommand === 'ui-safety-gate') {
cmdUiSafetyGate(cwd, args, raw);
return;
}
if (subcommand === 'verify-schema-drift') {
// Delegates to verify.schema-drift — drift capability gate at execute:wave:post (blocking).
// Dot-to-hyphen normalization means query "verify.schema-drift" routes here.
// Honor GSD_SKIP_SCHEMA_CHECK=true to bypass the gate (preserves the original inline gate behavior).
const phaseArg = typeof args[2] === 'string' ? args[2] : '';
const skipSchemaCheck = process.env['GSD_SKIP_SCHEMA_CHECK'] === 'true';
cmdVerifySchemaDrift(cwd, phaseArg, skipSchemaCheck, raw);
return;
}
if (subcommand === 'verify-codebase-drift') {
// Delegates to verify.codebase-drift — drift capability gate at execute:wave:post (non-blocking).
// Dot-to-hyphen normalization means query "verify.codebase-drift" routes here.
cmdVerifyCodebaseDrift(cwd, raw);
return;
}
if (subcommand === 'predicate') {
// Generic gate-predicate evaluator (#2008). The workflow gate-dispatch calls
// this for any gate whose `check` carries a `predicate` (instead of a `query`),
// passing the predicate object as --predicate '<json>'. NOTE: unlike the
// `check.query` subcommands above (which take positional phase args), this
// subcommand parses --flag value pairs.
cmdCheckPredicate(cwd, args, raw);
return;
}
if (subcommand === 'prohibition-enforcement') {
// The deterministic test-tier prohibition PRODUCER/gate (#1259, ADR-550 D5d). Locates the
// wired mechanical check (node-test or lint-rule), confirms fail-first, runs it, builds
// enforcementEvidence, and emits the dispositionForProhibition verdict. Invocable as
// `gsd_run check prohibition-enforcement <request.json>`.
routeProhibitionEnforcement(args, raw);
return;
}
error('Unknown check subcommand. Available: api-coverage-verify-pre, auto-mode, decision-coverage-plan, decision-coverage-verify, gap-analysis-plan-post, predicate, prohibition-enforcement, tdd-review-checkpoint, ui-plan-gate, ui-safety-gate, verify-schema-drift, verify-codebase-drift', ERROR_REASON.SDK_UNKNOWN_COMMAND);
}
export = {
routeCheckCommand,
decisionMentioned,
extractPlanDesignatedSections,
computeUiPlanGate,
computeUiSafetyGate,
cmdGapAnalysisPlanPost,
cmdTddReviewCheckpoint,
cmdCheckPredicate,
buildPredicateDeps,
parsePredicateFlags,
// Fail-closed phase-scope reader for the api-coverage gate — exported for
// in-process failure-injection tests (#2365 review).
readPhaseScope,
};