Skip to content

fix: 🐛 small follow-ups from the reviews of the leak checks, adoption docs and skill - #36

Merged
robertodr merged 5 commits into
mainfrom
fix/review-follow-ups
Nov 5, 2026
Merged

robertodr merged 5 commits into
mainfrom
fix/review-follow-ups

Conversation

@robertodr

Copy link
Copy Markdown
Member

🤖 AI text below 🤖

Follow-ups of the reviews of #26–#29.

Leak checks

  • scripts/check_internal_names.py: paths git quotes (spaces, a double quote, non-ASCII) are reported as named; output of gh that is not the listing skips the check instead of crashing the hook; an empty listing says "nothing to check" instead of "skipped".
  • tests/test_leaks.py: each tracked file is read once for both repository-wide tests, and a tracked file deleted but not staged is left out.

docs/repository-setup.md, step 2

  • The warning about tag-push workflows covers tagging a published version, not only v0.0.0.
  • The history snippet runs in a subshell that removes its temporary directory and the worktree however it ends.

skills/adopt-release-tools/SKILL.md

  • gh release list --limit 1000; inlined CodeArtifact values compared with their variables in the facts table; the old-process docs row widened to any manual step; the tag-push row covers tagging a published version.
  • Rulesets with enforcement disabled or evaluate stay out of step 8; an organisation ruleset handed to an owner has a done state.
  • branches-ignore named for the release/* checks; REPO set in the handed-over credentials command; command substitution counts as a heredoc; every report ends with the commands run that were not reads.

The public-visibility target and the payload-path note were already in place on main.

Closes #31

robertodr and others added 5 commits November 5, 2026 15:05
… from no access

- A path git quotes (a double quote, a backslash, a control character) is
  unquoted, non-ASCII paths are no longer escaped, and the tab git appends
  to a path with a space in a diff header is dropped.
- Output of gh that is not the expected listing skips the check, like a gh
  that fails, instead of crashing the hook.
- When gh lists no non-public repositories the hook says so, rather than
  that it was skipped.

Refs #31

Assisted-by: Pi:claude-opus-5-5
…rking tree

Refs #31

Assisted-by: Pi:claude-opus-5-5
… up after the history snippet

- The warning about workflows that publish on tag pushes covers tagging a
  published version too, not only v0.0.0.
- The snippet that generates the sections before the baseline runs in a
  subshell that removes its temporary directory and the worktree even when
  git-cliff fails.

Refs #31

Assisted-by: Pi:claude-opus-5-5
- Read every release, not the first 30.
- Compare each inlined CodeArtifact value with its variable in the facts
  table already.
- Widen the old-process docs to any manual step, and the tag-push check to
  tagging a published version.
- Leave rulesets with enforcement disabled or evaluate out of step 8, and
  give an organisation ruleset handed to an owner a done state.
- Name branches-ignore for the release/* checks, set REPO in the handed-over
  credentials command, count command substitution as a heredoc, and end
  every report with the commands that were not reads.

Refs #31

Assisted-by: Pi:claude-opus-5-5
@robertodr
robertodr merged commit 974c270 into main Nov 5, 2026
5 checks passed
@robertodr
robertodr deleted the fix/review-follow-ups branch November 5, 2026 15:34
@algorithmiq-release-bot

Copy link
Copy Markdown
Contributor

This fix isn't in a release yet. To ship it in v1.3.1, add the backport label; it works after merging too.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Small follow-ups from the reviews of the leak checks, CodeArtifact stub, adoption docs and skill

1 participant