Repository navigation
Converge 373 public stars and schedule weekly review intake - #978
Conversation
seathatflowsinourveins
added
the
lane:foundation
|
Dependency limit exceeded — report not shown. This pull request scan exceeded the 10,000-dependency limit applied to this scan, so the results are incomplete and may be inaccurate. To avoid reporting false positives, Socket has not posted a report. Upgrade your plan to raise the dependency limit and get complete reports, or view the partial scan in the dashboard. Socket is always free for open source. If this is a non-commercial open source project, contact us to request a free Team account. |
5831021 to
c7a4603
Compare
|
@codex review |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6b91bb7166
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
6b91bb7 to
3bbc426
Compare
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 3bbc426cab
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 2b5c9abe0d
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: f1dd19a423
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: f1dd19a423
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Follow jdx/mise v2026.10.3 (33a839a4d19acca617bc9ec0fe53f61266d60a9d), docs/dev-tools/shims.md and src/cli/activate.rs. The sibling PATH omitted the installed gh shim; the prior fixture masked that failure. Regressions fail first at c7a4603 without a fake gh, then pass with the rendered shim path and real installed CLI. Report a missing gh explicitly, cite CPython v3.13.16 subprocess exceptions, and retain fresh snapshots and notifications on failure. All 552 touched-module tests and FULL validate pass.
f1dd19a to
91a7955
Compare
…'s new reads CI (validate-shard-2) failed tests.test_runtime_worker_openhands_push_gate.RepositoryWorkflowTests.test_advisory_baseline_and_explicit_categories_stay_bounded_on_this_repository: it compares the advisory unclassified reads the gate derives from the HEAD tree with blueprints/runtime-workers/openhands/evidence/unclassified-gate-reads-20261004.json. The Claude-link check of 2bcd7e3 adds two reads in tools/adoption/install_skills.py (lines 315 and 317; the five earlier ones move to lines 464, 465, 673, 700 and 701), so the derivation counts 334 where the snapshot says 332. The snapshot is regenerated by the repository's own derivation (GitTree(ROOT, "HEAD") and derive_ci_protected, as the test runs it): count 332 to 334, the installer 5 to 7 locations, the shapes per script unchanged, no other script differs. The provenance keys are set as the last refresh (#978) set them: base_commit is this pull request's merge base, previous_snapshot_sha256 the sha256 of the file before, source_files_sha256 the installer at this head; the date is 2026-10-11; superseded_candidate_snapshot_sha256 named a candidate of #978's refresh and is dropped. The failing test, the push-gate module and the two OpenHands modules (389 tests) pass; the snapshot is registered in manifests/evidence.json. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
1 participant
Scope
Converge the public-stars catalog to 373 explicit scope dispositions, link all 315 in-scope repositories to 535 layer seeds, and provide a weekly external review queue through the existing GitHub CLI and native systemd timer. The five additions receive pinned primary-source reviews; the other 368 repositories retain explicit metadata and historical-seed evidence. Tier verdicts remain separate recorded follow-up.
The correction preserves unresolved review work through consecutive snapshot/journal publication failures. One atomic prepared record contains both documents; retry publishes the saved snapshot and journal before notification or another fetch. First-observed dates survive recovery, and both reopened work and later empty-queue clearances have notification identities that deduplicate their retries.
The decision index registers the canonical current-star snapshot and distinguishes five pinned source reviews from 368 metadata convergence references. Every dynamically resolved convergence document participates in the explorer's input digest, and seed source objects must declare the same repository. Existing timestamp, tier-follow-up and private-field checks remain enforced. Historical seeds retain
metadata_seed_reviewand the byte-identical master capture; four cross-repository React pointers were removed from the awesome-react seeds. The maintenance README documents the actual installed-CLI check.7b76a88c3e669dac328a8e05018ef0ccc029fd8f.f1dd19a423888a1f5da6221ae65120d610117753.lane:foundation.The terminal correction changes exactly two files: the native monitoring record's public_starred.py locators from :108/:450 to :109/:486, and that record's SHA-256 in evidence.json. Count 332, shape keys, other metadata and monitoring_only policy are unchanged. The native GitTree/derive_ci_protected comparison fails at 2b5c9ab and passes at this correction head; FULL validation and evidence ordering pass. Regression and producer-citation proof are retained in the external correction packet, keeping this commit within the CC's two-file scope.
SOTA sources
--paginate --slurp.src/cli/activate.rs:63–72anddocs/dev-tools/shims.md:71–80, the noninteractive shim PATH.6b91bb7166280a544280cd35f6961ccb4c571e97.catalogs/convergence-practice/sources/master-seeds-20261010.json, SHA-2564a62af4ca834d6170e24d511f4982dc864d1b15502534733cd3fe9cf2cc633ad, 3,554 strings and 528 checked joins. This is discovery evidence, with no claim of pinned upstream review for all older repositories.The review packet's
sources/MANIFEST.sha256binds 39 retained captures: the five new repositories' source files and implementation sources, eight verified release tags, the original new-five metadata/default-head record, master seed, and the baseline index/audit/templates. All twelve new-repository file hashes and Git blob identities match the pinned upstream bytes. The original raw public API response was deliberately not retained; acquisition evidence is the allowlisted catalog and its native writer reproduction. Transitive historical citations embedded in the master are not all retained or independently verified. The new correction packet retains the pinned CPython atomic-replacement source and four exact pre-fix repository modules under its ownsources/MANIFEST.sha256.Evidence-class table
source_reviewseeds and 528metadata_seed_reviewseedsLive weekly execution remains untested. No host unit was installed, enabled or started. Metadata/seed inclusion and scope review make no adoption, comparative SOTA, benchmark or runtime-acceptance claim.
Local commands run
Module runs used
PYTHONDONTWRITEBYTECODE=1 PYTHONPATH=. python3 -B -m unittest <module>; slower runs usednice -n 10 ionice -c2 -n7. Passing modules: tests.test_public_starred (34), tests.test_starred_convergence (27), tests.test_catalog_decisions (17), tests.test_starred_maintenance (14), tests.test_ecosystem_manifest (87), tests.test_validate (86), tests.test_local_pages_policy_grants (21), tests.test_stack_lifecycle (6), tests.test_verdict_review_gate (173), and tests.test_runtime_worker_openhands_push_gate (125).The publication correction at 2b5c9ab follows 3bbc426; exactly one terminal locator correction follows 2b5c9ab. Earlier branch commits remain unchanged. Current main was read at 78566b3. Native merge-tree verification of the complete staged source against that main is clean, so this correction required no rebase. The union retains 901 repositories, with 373 canonical current stars and 528 beyond current stars. Grants remain 92; the native monitoring check over the final source retains 332 entries. The publication correction updated 14 existing evidence rows; the terminal correction updates only the monitoring record's hash row, adds/removes none, and retains all 11,493 files.
The external co-op correction packet
pr978-journal-recovery-20261011retains the receipt, checksum-bound sources and native maintenance module's full verbose passing log, including both real systemd cases. The atomic-replacement citation also matches the pinned local mirrorpython/cpython@cbc944f4bc59639a444dd971c737788ba2283a91,Doc/library/os.rst:2754-2760.Follow-ups
WS1 stars-finalization follow-up (Codex 4242466476): require every seed row to be referenced by a convergence disposition; include a fail-first test (CC adjudication 18:24Z).
WS1 stars-finalization follow-up (Codex 4242466473): make a migrated canonical snapshot that loses convergence_manifest fail closed rather than enter legacy compatibility; include a fail-first test (CC adjudication 18:24Z).
WS1 stars-finalization follow-up (Codex 4242466470): check every qualifying source_review seed URL's owner/repository against the normalized seed repository; include a fail-first test (CC adjudication 18:24Z).
Normalize descriptive PENDING prefixes consistently for age checks and queue retention (Codex 4242307043); behavior dates from the first commit and current data has zero PENDING dispositions.
Scan the directly loaded verdict REGISTRY for private fields (Codex 4242307025); behavior is pre-existing at 3bbc426 and the committed registry has zero prohibited keys.
Fix README:84's absent-unit list-unit-files preflight in the next WS1 stars-finalization PR (Codex 4242307034); the CC will use the corrected check at post-landing installation.
Validate review_level against the seeds' evidence_class; Codex 4241889174 is pre-existing, while the current five source-review and 368 metadata-review rows are consistent.
Resolve tier verdict_refs through the recorded REGISTRY rather than accepting any matching local object; Codex 4241889168 is pre-existing and committed rows currently have no verdict_refs.
Bind convergence evidence_refs to the row's repository; Codex 4241889171's committed enaqx/awesome-react pointers resolve to react/react (CC-adjudicated pre-existing follow-up).
Clarify or refresh the monitoring snapshot's historical source_files_sha256 metadata; both stored hashes already differ at the merge base and previous heads, so this adjacent issue remains outside the locator correction.
Give every pending-queue membership change a persisted transition identity; returning to an earlier nonempty queue can suppress its updated notification (pre-existing GPT confirmation follow-up).
Capture and independently verify the transitive historical sources cited inside the master-seed witness; its retained strings and joins establish discovery provenance.
Measure live weekly timer execution and notification delivery at the post-landing CC installation cue.
Decision record
docs/decisions/2026-10-10-starred-repository-convergence.mdrecords mechanisms, pinned sources, boundaries, corrections and the replacement condition.Host evidence
No evidence/hosts receipts changed.
Checklist