fix(#4267): correct tdd.md pointer citations; fix(#4269): collapse plan-level gate-rule duplication (#4295)

* fix(#4267): correct tdd.md pointer citations; fix(#4269): collapse plan-level gate-rule duplication

execute-plan.md and gsd-executor.md each cited "Red-Green-Refactor Cycle"
for three facts (commit-scope contract, fail-fast rule, error handling),
but only the commit-scope contract lives there. Fail-fast is in tdd.md's
"Fail-Fast Rules" subsection (under "Gate Enforcement Rules") and error
handling is in tdd.md's "Error Handling" section — cite each correctly.

gsd-executor.md's "Plan-Level TDD Gate Enforcement" section also fully
restated the gate-sequence rules tdd.md's "Gate Enforcement Rules" already
owns (and covers more thoroughly, including the actual git-log validation
script). Collapse it to a short pointer, matching the treatment already
used by the cycle-steps pointer immediately above it.

Adds tests/tdd-reference-correctness.test.cjs asserting the pointer text
cites the correct section names, that those sections actually carry the
guidance, and that the old gate-sequence restatement is gone from
gsd-executor.md.

Closes #4267
Closes #4269

* docs(#4267): add changeset for tdd.md pointer correctness fix

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

* chore(#4267): acknowledge execute-plan.md growth

execute-plan.md grew 95 bytes (39766 -> 39861) from the corrected
three-section citation in the #3990/#4267 cycle-steps pointer.

Emitted-Drift-Ack-Growth: execute-plan.md — net +95 bytes from citing the "Fail-Fast Rules" and "Error Handling" sections by name instead of a single mis-scoped section (#4267).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

* fix(#4267): update PROSE_ALLOWLIST line numbers shifted by the pointer-citation edit

This branch's edits to agents/gsd-executor.md and gsd-core/workflows/execute-plan.md
shifted line numbers, leaving tests/no-bare-gsd-tools-command-position.test.cjs's
PROSE_ALLOWLIST pointing at stale lines. Update both entries to their new correct
lines (811 and 419 respectively) without changing the underlying prose.

* fix(#4267): restore INVALID_RED citation, fix allowlist line shift after #3770 rebase

The rebase onto next picked up #3770's already-merged fail-fast update to
gsd-executor.md's plan-level gate section, which this branch's own commit
collapses into a pointer. The conflict resolution kept the pointer but
dropped the literal "INVALID_RED" term that tests/tdd-red-evidence.test.cjs
requires gsd-executor.md to name — restored it. Also updates
no-bare-gsd-tools-command-position.test.cjs's PROSE_ALLOWLIST line number
for gsd-executor.md, shifted again by the rebase.

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

* test(#4267): fix non-matching regression-guard regex in tdd-reference-correctness

The fail-fast regression guard asserted gsd-executor.md no longer contains
"If a test passes unexpectedly during the RED phase" — but the actual old
prose (removed by this branch's pointer-collapse) read "If a test passes
unexpectedly during RED, STOP". The regex never matched the real old text,
so the assertion would have passed even against the unmodified pre-change
file. Caught by an isolated orthogonal review pass. Fixed to match the
actual removed wording, and confirmed (via a direct grep) it is genuinely
absent from the current file.

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

* docs(#4267): backfill changeset PR number

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

---------

Co-authored-by: sim <sim@local>
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
Tom Boucher
2026-09-04 19:13:56 -04:00
committed by GitHub
parent 2e056488d9
commit 5214ad5802
5 changed files with 146 additions and 23 deletions

View File

@@ -0,0 +1,5 @@
---
type: Fixed
pr: 4295
---
**TDD dispatch pointers now cite the tdd.md sections that actually carry the material they promise** — the RED/GREEN/REFACTOR pointer in both `execute-plan.md` and `agents/gsd-executor.md` cited a single section for the commit-scope contract, fail-fast rule, and error handling, but only the commit-scope contract lived there; each is now cited correctly. `agents/gsd-executor.md`'s plan-level gate-enforcement rules, previously restated in full alongside `tdd.md`'s own copy, now point at `tdd.md` as the single owner. (#4267, #4269)

View File

@@ -399,23 +399,19 @@ When executing task with `tdd="true"`:
**1. Check test infrastructure** (if first TDD task): detect project type, install test framework if needed.
**2-4. RED → GREEN → REFACTOR (#3990: stated ONCE):** execute the cycle exactly as the
canonical `gsd-core/references/tdd.md` "Red-Green-Refactor Cycle" section specifies (embedded
when TDD applies) — its commit-scope contract, fail-fast rule, and error handling. The
reference is the single source; do not improvise a variant.
**2-4. RED → GREEN → REFACTOR (#3990: stated ONCE; #4267: cited correctly):** execute the
cycle exactly as the canonical `gsd-core/references/tdd.md` reference specifies (embedded when
TDD applies) — the "Red-Green-Refactor Cycle" section's commit-scope contract, the "Gate
Enforcement Rules" section's "Fail-Fast Rules" subsection, and the "Error Handling" section.
The reference is the single source; do not improvise a variant.
## Plan-Level TDD Gate Enforcement (type: tdd plans)
## Plan-Level TDD Gate Enforcement (type: tdd plans, #4269: stated ONCE)
When the plan frontmatter has `type: tdd`, the entire plan follows the RED/GREEN/REFACTOR cycle as a single feature. Gate sequence is mandatory:
**Fail-fast rules (#3770):** If a test passes unexpectedly during RED, STOP — do NOT skip RED. A nonzero exit alone is NOT RED either: persist the RED evidence (command, exit code, failing test, expected, actual) and verify with `gsd_run check tdd-red-evidence <record.json>` — only `RED_EVIDENCE_OK` (the TARGET test failed an assertion for the behavior) authorizes GREEN; any INVALID_RED (zero tests, fixture crash, parser error, wrong test) blocks it.
**Gate sequence validation:** After completing the plan, verify in git log:
1. A `test(...)` commit exists (RED gate)
2. A `feat(...)` commit exists after it (GREEN gate)
3. Optionally a `refactor(...)` commit exists after GREEN (REFACTOR gate)
If RED or GREEN gate commits are missing, add a warning to SUMMARY.md under a `## TDD Gate Compliance` section.
When the plan frontmatter has `type: tdd`, the mandatory RED/GREEN/REFACTOR gate sequence,
its fail-fast rules (including the #3770 INVALID_RED / intentional-RED-evidence requirement
enforced via `gsd_run check tdd-red-evidence`), and the `## TDD Gate Compliance` SUMMARY.md contract are
specified in the canonical `gsd-core/references/tdd.md` "Gate Enforcement Rules" section
(embedded when TDD applies). The reference is the single source; do not improvise a variant.
</tdd_execution>
## MVP+TDD Gate

View File

@@ -292,12 +292,13 @@ End with: **Total deviations:** N auto-fixed (breakdown). **Impact:** assessment
For `type: tdd` plans — RED-GREEN-REFACTOR:
1. **Infrastructure** (first TDD plan only): detect project, install framework, config, verify empty suite
2. **Cycle (#3990: stated ONCE):** execute RED → GREEN → REFACTOR exactly as specified in the
canonical `~/.claude/gsd-core/references/tdd.md` "Red-Green-Refactor Cycle" section — its
commit-scope contract (`test({phase}-{plan})` → `feat({phase}-{plan})` →
`refactor({phase}-{plan})`, RED must fail, GREEN must pass, REFACTOR commits only on change),
its fail-fast rule, and its error handling. The reference is the single source; do not
improvise a variant.
2. **Cycle (#3990: stated ONCE; #4267: cited correctly):** execute RED → GREEN → REFACTOR
exactly as specified in the canonical `~/.claude/gsd-core/references/tdd.md` reference — the
"Red-Green-Refactor Cycle" section's commit-scope contract (`test({phase}-{plan})` →
`feat({phase}-{plan})` → `refactor({phase}-{plan})`, RED must fail, GREEN must pass, REFACTOR
commits only on change), the "Gate Enforcement Rules" section's "Fail-Fast Rules" subsection,
and the "Error Handling" section. The reference is the single source; do not improvise a
variant.
</tdd_plan_execution>
<precommit_failure_handling>

View File

@@ -103,11 +103,11 @@ const BARE_COMMAND_RE = new RegExp(
// Each entry MUST carry a one-line reason; the test prints the allowlist on
// failure so a reviewer can see exactly what is sanctioned.
const PROSE_ALLOWLIST = [
{ file: 'agents/gsd-executor.md', line: 816, reason: 'describes the SDK return envelope of `gsd-tools query commit`; not an instruction to run the bare word' },
{ file: 'agents/gsd-executor.md', line: 812, reason: 'describes the SDK return envelope of `gsd-tools query commit`; not an instruction to run the bare word' },
{ file: 'agents/gsd-phase-researcher.md', line: 33, reason: 'package-legitimacy provenance rule names the command as the source of an OK verdict; descriptive' },
{ file: 'agents/gsd-roadmapper.md', line: 647, reason: 'parenthetical "e.g." naming SDK queries a user *could* run; not an agent instruction' },
{ file: 'agents/gsd-intel-updater.md', line: 40, reason: 'cross-platform note names the `gsd-tools intel <subcommand>` CLI surface descriptively ("CLI invocations go through..."); not an agent instruction' },
{ file: 'gsd-core/workflows/execute-plan.md', line: 418, reason: 'describes the downstream SDK validation step (`validated downstream by ...`); names the mechanism, does not instruct the agent to type it' },
{ file: 'gsd-core/workflows/execute-plan.md', line: 419, reason: 'describes the downstream SDK validation step (`validated downstream by ...`); names the mechanism, does not instruct the agent to type it' },
];
// Resolver-snippet definition lines / probes that must never be flagged. A line

View File

@@ -0,0 +1,121 @@
'use strict';
/**
* #4267 / #4269 — tdd.md pointer citations must name the section that
* actually carries the guidance, and gsd-executor.md's plan-level gate
* restatement collapses to a pointer at tdd.md's "Gate Enforcement Rules".
*
* #4228 introduced pointers in execute-plan.md and gsd-executor.md that all
* cite a single section — "Red-Green-Refactor Cycle" — for three distinct
* pieces of guidance (commit-scope contract, fail-fast rule, error handling).
* Only the commit-scope contract actually lives there; the fail-fast rule is
* in "Fail-Fast Rules" (a subsection of "Gate Enforcement Rules") and error
* handling is in its own "Error Handling" section. (#4267)
*
* Separately, gsd-executor.md's "## Plan-Level TDD Gate Enforcement" section
* fully restates gate-sequence rules that tdd.md's "## Gate Enforcement
* Rules" section already owns (and covers more thoroughly, including the
* actual bash validation script). (#4269)
*/
const { test, describe } = require('node:test');
const assert = require('node:assert/strict');
const fs = require('node:fs');
const path = require('node:path');
const ROOT = path.join(__dirname, '..');
const read = (p) => fs.readFileSync(path.join(ROOT, p), 'utf8');
describe('#4267 — pointer text cites the correct tdd.md section per fact', () => {
test("execute-plan.md's cycle pointer cites three distinct, correctly-named sections", () => {
const plan = read('gsd-core/workflows/execute-plan.md');
const pointer = plan.slice(plan.indexOf('<tdd_plan_execution>'), plan.indexOf('</tdd_plan_execution>'));
assert.ok(/"Red-Green-Refactor Cycle"/.test(pointer),
'commit-scope contract must cite "Red-Green-Refactor Cycle"');
assert.ok(/"Fail-Fast Rules"/.test(pointer),
'fail-fast rule must cite "Fail-Fast Rules" by its literal section name, not just the words "fail-fast"');
assert.ok(/"Error Handling"/.test(pointer),
'error handling must cite "Error Handling" by its literal section name');
});
test("gsd-executor.md's cycle pointer cites three distinct, correctly-named sections", () => {
const executor = read('agents/gsd-executor.md');
const pointer = executor.slice(executor.indexOf('<tdd_execution>'), executor.indexOf('## Plan-Level') > -1 ? executor.indexOf('## Plan-Level') : executor.indexOf('</tdd_execution>'));
assert.ok(/"Red-Green-Refactor Cycle"/.test(pointer),
'commit-scope contract must cite "Red-Green-Refactor Cycle"');
assert.ok(/"Fail-Fast Rules"/.test(pointer),
'fail-fast rule must cite "Fail-Fast Rules" by its literal section name, not just the words "fail-fast"');
assert.ok(/"Error Handling"/.test(pointer),
'error handling must cite "Error Handling" by its literal section name');
});
test('tdd.md\'s "Fail-Fast Rules" section actually contains fail-fast guidance', () => {
const tdd = read('gsd-core/references/tdd.md');
const start = tdd.indexOf('### Fail-Fast Rules');
assert.ok(start !== -1, 'tdd.md must carry a "### Fail-Fast Rules" section');
const end = tdd.indexOf('### Executor Gate Validation', start);
const section = tdd.slice(start, end === -1 ? undefined : end);
assert.ok(/Unexpected GREEN in RED phase/i.test(section),
'the Fail-Fast Rules section must actually describe the unexpected-pass-in-RED rule');
});
test('tdd.md\'s "Error Handling" section actually contains error-handling guidance', () => {
const tdd = read('gsd-core/references/tdd.md');
const start = tdd.indexOf('## Error Handling');
assert.ok(start !== -1, 'tdd.md must carry an "## Error Handling" section');
const end = tdd.indexOf('## Commit Pattern for TDD Plans', start);
const section = tdd.slice(start, end === -1 ? undefined : end);
assert.ok(/fail/i.test(section) && /RED|GREEN|REFACTOR/.test(section),
'the Error Handling section must actually describe fail-in-RED/GREEN/REFACTOR guidance');
});
});
describe('#4269 — plan-level gate rules are owned once, by tdd.md', () => {
// The OLD gsd-executor.md restatement's distinctive two-line gate-sequence
// pattern (its own phrasing, distinct from tdd.md's).
const OLD_RESTATEMENT_MARKERS = [
/A `test\(\.\.\.\)` commit exists \(RED gate\)/,
/A `feat\(\.\.\.\)` commit exists after it \(GREEN gate\)/,
];
test('gsd-executor.md no longer restates the plan-level gate rules inline', () => {
const executor = read('agents/gsd-executor.md');
for (const marker of OLD_RESTATEMENT_MARKERS) {
assert.ok(!marker.test(executor),
`gsd-executor.md must not restate the old gate-sequence wording ${marker} — point at tdd.md's "Gate Enforcement Rules" instead`);
}
assert.ok(/references\/tdd\.md/.test(executor) && /Gate Enforcement Rules/.test(executor),
'gsd-executor.md must point at tdd.md\'s "Gate Enforcement Rules" section');
});
test('gsd-executor.md\'s plan-level gate section is now a short pointer, not a restatement', () => {
const executor = read('agents/gsd-executor.md');
const headingIdx = executor.search(/## Plan-Level TDD Gate Enforcement/);
assert.ok(headingIdx !== -1, 'the "## Plan-Level TDD Gate Enforcement" heading must still exist as an anchor');
const nextHeadingIdx = executor.indexOf('</tdd_execution>', headingIdx);
const section = executor.slice(headingIdx, nextHeadingIdx === -1 ? undefined : nextHeadingIdx);
const lineCount = section.split('\n').filter((l) => l.trim().length > 0).length;
assert.ok(lineCount <= 6,
`the plan-level gate section must be a short pointer (<=6 non-blank lines), got ${lineCount} lines:\n${section}`);
});
test('the gate-sequence rules (fail-fast + git-log validation + TDD Gate Compliance) appear in full only in tdd.md', () => {
const tdd = read('gsd-core/references/tdd.md');
const executor = read('agents/gsd-executor.md');
// tdd.md owns the full contract in its own words.
assert.ok(/Unexpected GREEN in RED phase/.test(tdd),
'tdd.md must carry the fail-fast rule');
assert.ok(/git log --oneline -E --grep/.test(tdd),
'tdd.md must carry the actual git-log gate validation script');
assert.ok(/## TDD Gate Compliance/.test(tdd),
'tdd.md must carry the TDD Gate Compliance SUMMARY.md contract');
// gsd-executor.md must not carry the actual validation script or the
// old fail-fast prose — only a pointer.
assert.ok(!/git log --oneline -E --grep/.test(executor),
'gsd-executor.md must not restate the git-log gate validation script — that lives only in tdd.md');
assert.ok(!/If a test passes unexpectedly during RED, STOP/.test(executor),
'gsd-executor.md must not restate the fail-fast rule prose — that lives only in tdd.md');
});
});