Repository navigation
chore: Re-apply the repo-ci-baseline Template - #263
Conversation
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
|
Copilot's review was requested for this PR. It is asked once per PR; if GitHub drops the request, review:claude calls in Claude instead. mpaulosky/dotfiles#360 |
There was a problem hiding this comment.
🟡 Changes recommended
The skills-list documentation still directs operators to pin the wrong repository and use an outdated skills path, causing skill setup to fail when followed.
1 open issue
What changed in this PR
Reapplies the repo-ci-baseline template updates to the repository’s CI workflows, Sandcastle automation, and branch policies.
Changes:
- Restricts published reviewer logs to bounded, fixed-message summaries and moves skill installation to Application Factory.
- Tightens Copilot bot identification and delays Claude fallback until the review-request record has settled.
- Allows pushes to Dependabot branches while excluding them from Sandcastle hand-offs, and removes
LICENSEfrom template-owned paths.
| File | Description |
|---|---|
scripts/tests/check-branch-name.test.sh |
Tests allowed and rejected Dependabot branch names. |
scripts/check-branch-name.sh |
Allows shell-safe Dependabot branches for pushes. |
.sandcastle/lib/role-logs.test.mts |
Tests fixed-message summaries, secret filtering, and log limits. |
.sandcastle/lib/role-logs.mts |
Publishes bounded summaries only for known build logs. |
.sandcastle/lib/review-ledger.test.mts |
Verifies that LICENSE is not template-owned. |
.sandcastle/lib/review-ledger.mts |
Removes LICENSE from template-owned paths. |
.sandcastle/lib/branches.test.mts |
Accounts for Dependabot’s exclusion from hand-offs. |
.sandcastle/lib/branches.mts |
Clarifies the existing hand-off restriction in a comment. |
.github/workflows/sandcastle.yml |
Installs skills from Application Factory and documents log sanitization. |
.github/workflows/pr-automerge.yml |
Tightens bot matching and waits before Claude fallback. |
.github/workflows/lint-actions.yml |
Updates the pinned zizmor action to v0.8.1. |
.github/scripts/tests/pr-automerge.test.mjs |
Adds bot-identity and fallback-timing regression coverage. |
.claude/rules/sandcastle-hand-off.md |
Documents that Dependabot branches cannot be handed off. |
Review effort: Balanced
Test Results Summary98 tests 98 ✅ 30s ⏱️ Results for commit 427f43e. ♻️ This comment has been updated with latest results. |
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #263 +/- ##
=======================================
Coverage 89.82% 89.82%
=======================================
Files 17 17
Lines 393 393
Branches 85 85
=======================================
Hits 353 353
Misses 15 15
Partials 25 25 🚀 New features to boost your workflow:
|
The re-Applied sandcastle.yml reads pinned skills from the Factory's main, not dotfiles. Comment lines only: the file still lists no pin and no skills. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Fixes R11 from the pre-push review of mpaulosky/TicketManager#263 (see #40). #37 made the Branch name check accept only `A-Za-z0-9._-` in a `dependabot/*` name, in both forms. The `--pr` form accepted any `dependabot/*` before, so a Dependabot PR whose branch carries another character (a scoped npm package's `@`, a version's `+`) would fail the required `Branch name` check. All 146 real Dependabot branch names in the Fleet match the pattern, but none is an ungrouped scoped npm package, so the data neither confirms nor rules it out. - `--pr` takes any `dependabot/*` again, as before #37; CI quotes the name. - The hook form stays shell-safe and now also takes `@` and `+`. - Tests: `@` and `+` in the first and later segments, `--pr` leniency, and refused `dependabot/`, `dependabot/a/`, `dependabot//a`, `Dependabot/x`, `claude/dependabot/x`, `xdependabot/a` and a name with a newline. Five mutations (`+` to `*` in each segment, dropping `@`/`+`, dropping the `--pr` branch, unanchoring) each fail a test. - `references/git-hooks.md` updated. `scripts/check-branch-name.sh` copied over; `scripts/gate.sh` and the Sandcastle check pass. Left in #40: R3-R9, R19 and the rest. Rollout: `scripts/check-branch-name.sh` and its test are Owned files, so this reaches the Fleet by a Rollout. TicketManager's re-Apply (#263) is waiting for it. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
…ry path (#42) Fixes R1, the Critical from the pre-push review of mpaulosky/TicketManager#263 (see #40). `askCopilotOnce` waited out the rest of `COPILOT_ASK_SETTLE_MS` (#23) only when the Copilot review request failed. Two other paths returned `true` about one poll after the request, with the ask record seconds old: the read-back's `catch`, and the `break` when the head moved. The caller then added `review:claude`; Claude Review skips a record younger than 30 s and a later PR Auto-Merge run strips the label, so Claude's review waited for another event or sweep. - One local `waitOutSettle()` (the same wait: the rest of the window, capped by `COPILOT_ASK_WAIT_MS`) is called on all three paths. A registered request needs none, and an unregistered one already waited the whole read-back window. - Tests on the mock clock: the read-back error and the moved head each hold the run, and `review:claude`, until the record is 30 s old; a registered request ends the run early. Both new tests fail without the fix. The existing failed-request test now has an upper bound, so removing the `COPILOT_ASK_WAIT_MS` cap fails it (the run took 30 s); this is R6's missing bound. A record already older than the window is still untested. - The header no longer says a failed request counts as dropped "at once" (R4). - The Factory's copy of `pr-automerge.yml` is copied over. `scripts/gate.sh` passes. Still in #40: R3, R5, R6 (the older-record case), R7, R8, R9, R19. Rollout: `pr-automerge.yml` and its test are Owned files, so this reaches the Fleet by a Rollout. TicketManager's re-Apply (#263) is waiting for it and then gets one review run covering both Template fixes. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
There was a problem hiding this comment.
🔵 Needs a closer look
Maintainer confirmation is needed that the intentionally retained channel in public reviewer-log artifacts (~72 selectable bits per log, up to 20 logs) is acceptable under the sandbox credential threat model.
No open issues 🎉
1 resolved since last review
Review effort: Balanced
Automated release blog posts for #263, opened by the release workflow. It holds every Release whose post isn't on main yet, rebuilt from main on each run. The [skip-release] title marker keeps its merge from starting another release. Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

Why
Re-Apply the repo-ci-baseline Template so this repo carries the Template changes from mpaulosky/Application_Factory#36, mpaulosky/Application_Factory#37, mpaulosky/Application_Factory#38 and mpaulosky/Application_Factory#39.
What changed
f1d2468Apply commit,apply.shoutput alone: brings chore(deps): Bump zizmorcore/zizmor-action from 0.8.0 to 0.8.1 in the minor-and-patch-actions group across 1 directory Application_Factory#36, chore(baseline): Let the hook accept dependabot/* so a person can fix a Dependabot PR Application_Factory#37, docs: Capitalize the commit examples in Instructions Application_Factory#38 and feat(baseline): Make LICENSE a Seed file so Apply keeps a project's year Application_Factory#39..claude/rules/sandcastle-hand-off.md.github/scripts/tests/pr-automerge.test.mjs.github/workflows/lint-actions.yml.github/workflows/pr-automerge.yml.github/workflows/sandcastle.yml.sandcastle/lib/branches.mts.sandcastle/lib/branches.test.mts.sandcastle/lib/review-ledger.mts.sandcastle/lib/review-ledger.test.mts.sandcastle/lib/role-logs.mts.sandcastle/lib/role-logs.test.mtsscripts/check-branch-name.shscripts/tests/check-branch-name.test.shVerification
The Apply commit is exactly
apply.sh's output on a branch up to date withmain.The pre-push gate (
scripts/gate.sh) passed onf1d2468.427f43eApply commit,apply.shoutput alone: brings chore(deps): Bump zizmorcore/zizmor-action from 0.8.0 to 0.8.1 in the minor-and-patch-actions group across 1 directory Application_Factory#36, chore(baseline): Let the hook accept dependabot/* so a person can fix a Dependabot PR Application_Factory#37, docs: Capitalize the commit examples in Instructions Application_Factory#38, feat(baseline): Make LICENSE a Seed file so Apply keeps a project's year Application_Factory#39, fix(baseline): Keep the --pr branch check lenient for Dependabot Application_Factory#41 and fix(baseline): Hold review:claude until the ask record settles on every path Application_Factory#42.🤖 Generated with Claude Code