golem15/msd-core
4
0
Fork 0
Code Issues 1 Pull Requests Actions 3 Packages Projects Releases Wiki Activity

fix(verification): keep passed uat results across verifier re-runs #2

Merged
jakub merged 2 commits from fix/uat-verification-stale-loop into next 2026-10-11 10:44:26 +00:00
Conversation 0 Commits 2 Files Changed 29 +42 -44
3 changed files with 42 additions and 44 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Showing only changes of commit 7b50026075 - Show all commits

18
tests/fixtures/compact-content-benchmark-baseline.json vendored
View File

@@ -18,9 +18,9 @@
"reductionPct": 16.74
},
"execute-phase": {
"offTokens": 24960,
"onTokens": 22854,
"reductionPct": 8.44
"offTokens": 25132,
"onTokens": 23026,
"reductionPct": 8.38
},
"new-project": {
"offTokens": 14038,
@@ -33,14 +33,14 @@
"reductionPct": 11.06
},
"verify-work": {
"offTokens": 13192,
"onTokens": 11223,
"reductionPct": 14.93
"offTokens": 13136,
"onTokens": 11167,
"reductionPct": 14.99
}
},
"aggregate": {
"offTokens": 107146,
"onTokens": 91020,
"reductionPct": 15.05
"offTokens": 107262,
"onTokens": 91136,
"reductionPct": 15.03
}
}

14
tests/progress-forensic.test.cjs
View File

@@ -209,7 +209,7 @@ describe('#1107: progress routing consults verification.status before reporting
);
});
test('stale verification routes to verify-work (Route V.stale)', () => {
test('stale verification is re-verified in place, as one action (Route V.stale)', () => {
const workflow = readWorkflow();
assert.ok(workflow.includes('**Route V.stale:'), 'must define a Route V.stale section');
const route = workflow.slice(
@@ -217,8 +217,16 @@ describe('#1107: progress routing consults verification.status before reporting
workflow.indexOf('**Route V.gaps:')
);
assert.ok(
route.includes('verify-work'),
'Route V.stale must route to /msd:verify-work {phase}'
route.includes('msd-core/references/stale-reverification.md'),
'Route V.stale must follow the shared stale re-verification procedure'
);
assert.ok(
route.includes('.reverify_phases'),
'Route V.stale must batch every stale-but-executed phase from init.progress, not just the current one'
);
assert.ok(
!/`\/msd:execute-phase \{phase\}/.test(route),
'Route V.stale must not send the user to execute-phase for a phase whose plans are all summarized'
);
});

54
tests/verify-work-auto-transition.test.cjs
View File

@@ -98,14 +98,18 @@ describe('verify-work.md — auto-transition after UAT passes with 0 issues', ()
test('human_needed verification is promoted to passed only after successful human UAT', () => {
const content = fs.readFileSync(VERIFY_WORK, 'utf-8');
const statusIdx = content.indexOf('VERIFICATION_STATUS=$(msd_run query verification.status "$PHASE_DIR"');
const humanNeededIdx = content.indexOf('if [ "$VERIFICATION_STATUS_VALUE" = "human_needed" ]; then');
const setPassedIdx = content.indexOf('msd_run query frontmatter.set "$VERIFICATION_FILE" --field status --value passed');
const staleIdx = content.indexOf('If `PHASE_VERIFICATION_STATUS` is `stale`');
const canonIdx = content.indexOf('CANON=$(msd_run query verification.canonicalize-uat "$PHASE_DIR"');
const predicateIdx = content.indexOf('PHASE_COMPLETE=$(msd_run phase uat-passed "{phase}" --require-verification)');
assert.ok(statusIdx !== -1, 'verify-work.md must inspect canonical verification status');
assert.ok(humanNeededIdx > statusIdx, 'status=passed promotion must be restricted to human_needed');
assert.ok(setPassedIdx > humanNeededIdx, 'human_needed verification must be promoted after status check');
assert.ok(setPassedIdx < predicateIdx, 'verification must be canonicalized before the required predicate runs');
assert.ok(canonIdx > statusIdx, 'the promotion must go through the shared canonicalize-uat seam, after the status read');
assert.ok(staleIdx < canonIdx, 'a stale report must be re-verified BEFORE canonicalization, or a passed UAT can never be recorded');
assert.ok(canonIdx < predicateIdx, 'verification must be canonicalized before the required predicate runs');
assert.ok(
!content.includes('frontmatter.set "$VERIFICATION_FILE" --field status --value passed'),
'the workflow must not hand-roll the status flip — verification.canonicalize-uat owns it'
);
});
test('stale verification blocks before phase transition', () => {
@@ -136,47 +140,33 @@ describe('verify-work.md — auto-transition after UAT passes with 0 issues', ()
// phase-close uses in its --uat-only form (the verification-status blocker
// is exactly what the flip removes, so the full predicate could never pass
// at pre-check time); the flagged call stays the later transition gate.
// The flip itself — gated on the uat-only row predicate, refused for any
// status other than human_needed, fail-closed on missing evidence — is owned
// by `verification.canonicalize-uat` and tested behaviorally in
// tests/uat-evidence.test.cjs. What stays pinned here is the workflow's side
// of that contract: it calls the seam, and it reports a refusal.
describe('verify-work.md — canonicalize flip is gated by the UAT predicate (#4663)', () => {
test('canonicalize flips to passed only when the uat-passed predicate reports passed (#4663)', () => {
test('the flagged predicate remains the later transition gate (#4663)', () => {
const content = fs.readFileSync(VERIFY_WORK, 'utf-8');
const humanNeededIdx = content.indexOf('if [ "$VERIFICATION_STATUS_VALUE" = "human_needed" ]; then');
const precheckIdx = content.indexOf('UAT_PRECHECK=$(msd_run phase uat-passed "{phase}" --uat-only 2>/dev/null)');
const flipGuardIdx = content.indexOf('if [ "$UAT_PRECHECK_PASSED" = "true" ]; then');
const setPassedIdx = content.indexOf('msd_run query frontmatter.set "$VERIFICATION_FILE" --field status --value passed');
assert.ok(precheckIdx !== -1, 'the canonicalize block must run the uat-passed predicate before flipping');
assert.ok(humanNeededIdx !== -1 && precheckIdx > humanNeededIdx, 'the pre-check must sit inside the human_needed branch');
assert.ok(content.includes(".passed // false"), 'the verdict must be extracted from the typed report with a false default');
assert.ok(flipGuardIdx !== -1 && flipGuardIdx > precheckIdx, 'the flip must be guarded on the extracted passed verdict');
assert.ok(setPassedIdx > flipGuardIdx, 'frontmatter.set must sit INSIDE the passed==true guard');
});
test('the canonicalize pre-check runs uat-passed without --require-verification (#4663)', () => {
const content = fs.readFileSync(VERIFY_WORK, 'utf-8');
const precheckIdx = content.indexOf('UAT_PRECHECK=$(msd_run phase uat-passed "{phase}" --uat-only 2>/dev/null)');
const canonIdx = content.indexOf('CANON=$(msd_run query verification.canonicalize-uat "$PHASE_DIR"');
const flaggedIdx = content.indexOf('PHASE_COMPLETE=$(msd_run phase uat-passed "{phase}" --require-verification)');
assert.ok(precheckIdx !== -1, 'the --uat-only pre-check must exist');
assert.ok(
!content.slice(precheckIdx, precheckIdx + 120).includes('--require-verification'),
'the pre-check is the unflagged predicate - requiring verification there would evaluate the very report being written'
);
assert.ok(flaggedIdx !== -1 && flaggedIdx > precheckIdx, 'the flagged predicate remains the later transition gate');
assert.ok(canonIdx !== -1, 'the canonicalize call must exist');
assert.ok(flaggedIdx !== -1 && flaggedIdx > canonIdx, 'the flagged predicate remains the later transition gate');
});
test('refused canonicalization says the verification stays human_needed (#4663)', () => {
const content = fs.readFileSync(VERIFY_WORK, 'utf-8');
assert.match(content, /"\$CANON_REASON" = "uat_not_passed"/, 'the refusal must branch on the typed reason');
assert.match(content, /stays human_needed/, 'the refusal message must say verification stays human_needed');
assert.match(content, /blockers \| length/, 'the refusal must carry the blocking-row count');
});
test('an indeterminate pre-check must not flip the report (fail closed) (#4663)', () => {
test('an indeterminate canonicalize call reads as no reason, never as a flip (#4663)', () => {
const content = fs.readFileSync(VERIFY_WORK, 'utf-8');
// jq -r '.passed // false' with an `|| echo "false"` fallback: empty or
// failed msd_run output must yield no-flip, never a flip.
assert.ok(
content.includes(`jq -r '.passed // false' 2>/dev/null || echo "false"`),
'the extraction must default to false on empty/failed output'
content.includes(`jq -r '.reason // empty' 2>/dev/null || echo ""`),
'the extraction must default to empty on empty/failed output'
);
});
});