Repository navigation
Conversation
…e epic archived Claude-Session: https://claude.ai/code/session_01D9yCR4qSsVzzoaHepAeTUW
…ase-intents, keep store-owns-claude-md-managed-block behind a re-baseline story Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
Nothing has read or written pm-suite-certification.json since 0.51.0 moved the record to pm-suite-certification.d/, and no pre-0.51.0 branch is live (branches: dev, main, rc/0.52.0). The pre-commit hook now removes a leftover copy with an idempotent rm -f beside the pm-isolation-flag cleanup; the directory is untouched. Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…rker value and the accepted set Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…al|work) and a defined work total Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
… — refresh the section whenever limit liveness changes Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…ns — name the change root as OpenSpec resolves it Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
… epic-writing verb and an epic.sessions trail Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…-to-feature map, commit-gate re-baseline story Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…mposition log in plain JSONL Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…switch only where the numbers justify it Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…eakdown, outlier flags, recommendations, trend Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…e them from the ledger Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
… file as pid 0
The stub wrote its ready file with writeFileSync (create, then fill), so under load the
guard saw the file created but empty, took Number("") = 0, and process.kill(0, 0) reported
its own process group alive for ever. The stub now renames the file into place and the
guard refuses a non-pid. Reproduced 1 in 40 concurrent runs under 24 CPU burners before,
0 in 60 after. Drops the entry from known-flakes.json.
Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
Twin-Unchanged: certify-index — test-side race fix only; the certify behaviour the twin covers is untouched
…do no liveness compare or refresh Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…stamp or refuse on an identity Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…ace id, amend-and-regate before a switch Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…profile maps to core Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…est tokens override governs Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
… the shared builder Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…, newest first Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
… Decisions section Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…opwatch The guard timed a spawned engine against the 2000 ms lock wait, so node's own start-up on a loaded machine counted as "waited out the lock". The engine returns the blocked refusal before its wait loop sleeps, and the wait-it-out refusal reads "was not released within"; asserting the message asserts the branch. 40 concurrent runs under 24 CPU burners: 0 failures. Empties known-flakes.json. Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM Twin-Unchanged: state-file-refuses-to-guess — test-side timing assertion replaced by a branch assertion; the engine behaviour the twin covers is untouched
… reading state.json (#204) Three emitted procedures need the set of existing epics (tracker-sync dedup, intake, a disposition's --deferral). The engine had no verb for it, so agents read .conductor/state.json and guessed its shape. `list [--status] [--lane] [--linked-to <id>] [--json]` filters the record; --linked-to is incoming. Read-only (verb-effects), command doc, parity-ledger entry, flag rows, and the sweep/recipe/baseline entries the registries demand. Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM Twin-Unchanged: conductor-31 — adds a VERB_BASELINE row and widens one closed-list entry for the new verb; the assertion twin reads the same registry and passed unchanged Twin-Unchanged: output-text-integrity — adds three POISON_RECIPES rows for the new verb's flags; the assertion twin's subject (escaping) is untouched Twin-Unchanged: verb-surface — adds one DISPATCH_BASELINE row for the new verb; the assertion twin sweeps the same dispatch table and passed unchanged
…e flake retry re-runs it node:test locates a failing test at the test() call, which for every unitTest() is unit-harness.mjs: a file the runner was not given, so failuresOf counted it unmapped and recoverFlakes never retried a unit-rung failure. The harness now puts a [unit-test-file: <path>] line at the top of the failure's stack (an AssertionError with a generated message prints only its stack), and failuresOf reads it back when the located file is not one it was given. A marker naming a file the runner was not given, or no marker, stays unmapped, so nothing that cannot be re-run is ever recovered. Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…ew-fail-requires-shas Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
Consolidates the 20 pending changesets into the 0.53.0 entry and removes them, plus 19 fragments already consolidated into 0.51.0 and 0.52.0 and never deleted. No MIGRATIONS entry: every field this release adds is optional. Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
…sons skills Covers list, log-detour --sha, retract-detour --minimal, the epic session trail, attributed-commits-hold-duplicates, --token-basis and the work total, named invalid spend markers, the rejected-lesson briefing line, and the live-limit spend-guard section. Corrects the claim that --session exists only on claim and unclaim. Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
|
Note Currently processing new changes in this PR. This may take a few minutes, please wait... ⚙️ Run configuration
📒 Files selected for processing: 224
✨ Finishing Touches
Comment |
The 1.3 capture tests failed on the CI leg that shifts Date.now() 400 days ahead: the fixture transcript kept its real mtime, so ingest skipped it as older than enabledAtMs. Claude-Session: https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
This was referenced Oct 21, 2026
Closed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Cuts pm 0.53.0 from
dev: 20 built epics plus the release cut. See the## [0.53.0]entry in CHANGELOG.md.--token-basis total|work, named invalid spend markers,spend-analysis,overhead,record-decision/decisions, spend-guard rules section that follows live limits.list,log-detour --sha,retract-detour, epic session trail, integrityattributed-commits-hold-duplicates, retired single-file certification record.known-flakes.jsonnow empty.set-dev-log/dev-log), off by default.Gate 2 recorded for all ten OpenSpec epics (one converged review, 0 Critical); release cross-spec review
pass(11 specs). Assertion half 2427/2427, functional 1359/1359, sweeps 25/25; full run 3811 tests, 0 failing. Mintlify site updated in a separate PR on the docs repo.https://claude.ai/code/session_01BoqzgrFwRC6QTUKuBkw8kM
Summary by CodeRabbit
New Features
Bug Fixes
Documentation
Release