Session Handoff [default]: KiCadLibrary Promoted and Released as 2.1.48, Build-Release-Asset Hook Proven #2319

Next steps, in priority order

Re-read each issue and pull request rather than trusting these summaries.

1. Finish KiCadLibrary's remaining standup items. This needs a fresh write grant for ptr727/KiCadLibrary, because the last grant covered that session only. Done means python3 spec/audit.py KiCadLibrary reports only the six secret defects and the version.json intent hint, and then a second promotion merged and released.
   • Re-carry the workflow contract. The hub's Record KiCadLibrary's Standup in the Registry and Rollout Trackers #2303 (e182edd) changed WORKFLOW.md and .github/skills/workflow-ci-contract/references/architecture.md after KiCadLibrary's Add devcontainer + per-OS host and SSH signing docs #63 carried them. The 2026-10-03T23:22:06Z audit reports the skill reference as matching a past hub revision and gives WORKFLOW.md an intent hint.
   • Wire the pre-commit framework with the doc gates and the ruff and mypy half.
   • Add the .vscode/tasks.json that the audit reports missing.
   • Clear ptr727/KiCadLibrary#62, the README and pointer cleanup.
2. Settle Order the LF Conversion Before Carrying LF Files Into a CRLF-Default Repository #2290 in one change. The comment on Order the LF Conversion Before Carrying LF Files Into a CRLF-Default Repository #2290 records every constraint from three review rounds, including the resync-a-repo and standup-a-repo skills that restate the apply order. Write all of it at once rather than patching finding by finding.
3. Hub follow-ups:
   • The Publisher Stub's Shared Concurrency Group Lets a No-Op Push Replace a Pending Dispatch #2300: the publisher stub's shared concurrency group.
   • Document the File-Target Caller Variant of the Release Chain and Drop the Branch-Matrix Trigger Shape #2301: the file-target caller variant docs and the stale branch-matrix sentence.
   • The Hooks Rollout Still Lists Blog as Doc-Gate Only #2302: the hooks rollout's Blog row.
   • A comments Label Applied by gh pr create Misses the Event That Starts the First Run #1703: the comments label race is still unfixed. This round opened Split Devcontainer and Workspace per Language and Drop Husky #65 unlabelled, then labeled it and closed and reopened it.
4. The other six repositories from Session Handoff [default]: Table Carry, Quota Safety and install-tools Shipped to Develop, Defined-Moments Review Gate #2278 Ready #2281: EspDinIoT, DevKitCIoT, HomeAssistant-Config, PhotoCleaner, AudioCleaner, and HolidayLights. Audit each one before choosing a procedure. EspDinIoT and DevKitCIoT are still on the CRLF default, so their LF conversion goes first, per the Order the LF Conversion Before Carrying LF Files Into a CRLF-Default Repository #2290 comment.
5. Carried from Session Handoff [default]: configure.sh Refuses Unknown Arguments, Released as 2.0.805 #2289: the rest of Every CLI Script Answers --help and Refuses Unknown Arguments Before Writing #2285, Audit Cross-Checks No Python Dependabot Ecosystem, and KiCadLibrary's requirements-dev.txt Is Untracked #2286's audit check, Run the Whole Test Suite Before a Push, Not the Modules a Change Touches #2280, Fix STANDUP.md's Prerequisite Item That Says Mechanisms Declare Environments #2270, and the items Session Handoff [default]: Table Carry, Quota Safety and install-tools Shipped to Develop, Defined-Moments Review Gate #2278 Ready #2281 carried from Session Handoff [default]: Fork Handoff Chains Released as 2.0.799, #2268 Fix on Develop #2273.

External blockers

• KiCadLibrary secrets. CODEGEN_APP_CLIENT_ID, CODEGEN_APP_PRIVATE_KEY, and CODECOV_TOKEN are missing from both its Actions and Dependabot stores. Only the maintainer can add them. Until then:
  • the merge-bot cannot auto-merge Dependabot pull requests there, so every KiCadLibrary release needs a dispatch
  • coverage uploads skip
  • the audit reports six secret defects.
• CodeRabbit on KiCadLibrary. The free plan allows one review per hour, and it was rate-limited for the whole of this round. Copilot was the only reviewer.
• A native Windows host. It is needed for the bootstrap.ps1 check, carried from Session Handoff [default]: WORKFLOW.md Corrections and the pyproject.toml Decision Released as 2.0.672 #1804.

Internal dependencies

• The build-release-asset hook is now proven, so the six-repository sweep no longer waits on KiCadLibrary. An eda repository can adopt the same shape.
• Step 2 before EspDinIoT and DevKitCIoT. The procedure fix lands before those two conversions, so they run against written rules rather than against the Order the LF Conversion Before Carrying LF Files Into a CRLF-Default Repository #2290 comment.
• From Session Handoff [default]: Table Carry, Quota Safety and install-tools Shipped to Develop, Defined-Moments Review Gate #2278 Ready #2281: Deny the Remaining core.hooksPath and no-verify Bypass Routes in the Guard #2177 follows or absorbs the auto-2112 heredoc work, and Decide How the Unattended Picker Recognizes an Adopted Lesson #2183 comes before Close the Unattended Loop's Parking-Comment and Interrupted-Park Gaps #2184.

State

• Hub:
  • main is unchanged at 2.0.807, 6b1b1477.
  • develop is one commit ahead with Deny a Signal-0 timeout as a Wait Bound in the Guard #2117 from the unattended auto-1634 lane, which merged after the last handoff was written. It is unpromoted.
  • There are no open hub pull requests. The primary checkout is on develop and clean, one commit behind origin/develop.
  • This session's detached hub worktree is removed at close. Re-derive the rest with git worktree list.
• KiCadLibrary:
  • main is at 0d02bfd, through promotion Split Devcontainer and Workspace per Language and Drop Husky #65, and released as 2.1.48. That release carries ptr727-KiCadLibrary-2.1.48.zip, built by the hub build-release-asset hook.
  • develop holds the same content after Add PyPiLibrary, Per-Language Devcontainers, Signing Docs, NuGet Rename #66.
  • Open pull requests are Dependabot's Develop #53 and Pin release action SHA, target_commitish, agent conventions #61 only. No feature branch or worktree remains.
  • The primary clone sits on main, 12 commits behind. It is the maintainer's surface and was not touched.
• Unattended lanes untouched: auto-2176, auto-2167, auto-2157, auto-2080, and python-contract. Re-derive with handoff.py tracks.
• Other repositories written: KiCadLibrary only:
  • pull requests Split Devcontainer and Workspace per Language and Drop Husky #65 and Add PyPiLibrary, Per-Language Devcontainers, Signing Docs, NuGet Rename #66
  • review-thread replies and resolves
  • a local-review attestation comment on Add PyPiLibrary, Per-Language Devcontainers, Signing Docs, NuGet Rename #66
  • the release dispatch
  • one close-and-reopen of Split Devcontainer and Workspace per Language and Drop Husky #65 to clear the label race.

The parked decision queue

Zero issues carry the decision label in ProjectTemplate or KiCadLibrary. This round's questions were asked as they came up and answered:

• the KiCadLibrary write grant, for this session and step 1 only
• merging Split Devcontainer and Workspace per Language and Drop Husky #65 although Copilot's file table covered 59 of its 97 changed files, and releasing it.

What the last round did

• ptr727/KiCadLibrary#65 merged and released as 2.1.48. It promoted Update codegen files #59, Update codegen files #60, Add devcontainer + per-OS host and SSH signing docs #63, Add PyPiLibrary Python sibling project #64, and Add PyPiLibrary, Per-Language Devcontainers, Signing Docs, NuGet Rename #66. This was the first real publish through the hub build-release-asset hook. The release zip holds the symbol library, the footprints, the 3D models, the library tables, and provenance.tsv.
• ptr727/KiCadLibrary#66 merged to develop. It fixed Copilot's Split Devcontainer and Workspace per Language and Drop Husky #65 finding, a fetch_docs.py docstring pointing at the README task list that Add PyPiLibrary Python sibling project #64 had moved to TODO.md. The class sweep found two more stale pointers, in scripts/README.md and in the CONVENTIONS.md unknown-source row.
• The promotion's coverage. Copilot round 1 on Split Devcontainer and Workspace per Language and Drop Husky #65 had a file table covering 59 of 97 files. The 38 it left out were 24 carried hub files (18 skills plus the governance docs) and 14 of the repository's own. Round 2 had no table. The maintainer accepted the partial coverage.
• Nothing was filed.

What not to repeat

• Skipping the dispatch because the merge started a publisher run. KiCadLibrary's publisher has a push-to-main trigger, so the promotion merge started a run. But publish-plan-task.yml publishes only on a bot merge or a dispatch, so that run planned and published nothing. The dispatch in merge-and-release step 5 is still owed. Read the plan job's publish= output before assuming a push run is the release.
• Elliptical wording in a pointer fix. "Tracked in TODO.md until it is" drew a Copilot finding. Name the condition outright.
• Running git push origin --delete from a primary checkout. The guard denies it and blocks the whole compound command with it. Delete through the API after checking the ref's object id, per the drive-pr verify-then-delete step.

New learnings

• Nothing new needed governance. The publisher dispatch point above is a slip against merge-and-release step 5, which already states the dispatch, so it stays here rather than becoming rule text.