Start a port when a porting issue is created

[ddaspit]

Quick summary

A new porting issue in this repo now starts Claude, which ports the change and opens a PR. A merged PR no longer gets a porting issue when it has the no porting label. No library code changes.

Where to look

• claude.yml: the porting label starts the workflow through the action's label_trigger. An issue created with a label fires a labeled event, so every new porting issue starts a port. The appended prompt tells Claude to use the port-pr skill. The author checks still apply. Porting issues are created with a member's token, so they pass them.
• claude.yml: the @claude mention clause skips labeled events and issues with the porting label. Otherwise any label on an issue that mentions @claude would run the full job for nothing, and a porting issue that mentions @claude would start two ports.
• create-porting-issue.yml: skips a merged PR with the no porting label. The existing check that skips a PR closing an auto-generated issue is still there, so a port does not create an issue to port it back.

Porting stays the default, so nothing needs to add a label to new PRs. I renamed the needs porting label I created earlier to no porting. No PR had it.

Deliberately not included

The porting issues in this repo are created by sillsdev/machine's copy of create-porting-issue.yml. The opt-out there needs the same change.

Validation

• Parsed claude.yml and create-porting-issue.yml with PyYAML. Both parsed.
• git diff --check -- clean.
• I did not run ./local_check.sh. This PR only changes workflow files.
• None of this can be verified until the workflows run after merge.

🤖 Generated with Claude Code

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.07%. Comparing base (8d3767c) to head (f3df62a).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #393   +/-   ##
=======================================
  Coverage   92.07%   92.07%           
=======================================
  Files         394      394           
  Lines       24896    24896           
=======================================
  Hits        22922    22922           
  Misses       1974     1974           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[claude]

1. Verdict: approve.
2. Nothing blocks. Since 5aae6d5, f3df62a turns porting into an opt-out. It deletes label-needs-porting.yml and flips the gate at .github/workflows/create-porting-issue.yml:14-16 to !contains(..., 'no porting'). Against main, the only change to that file is the opt-out clause. Every merged PR without the label still gets an issue, as it did on main.
3. Critical 0, Important 0, Low 0 new. F1 (Low) was already addressed.
4. What I ran:
   • compare 5aae6d5...f3df62a returned ahead.
   • git diff 5aae6d5 f3df62a: the gate flip plus the deleted labeling workflow. pull_request_target is gone from the PR, so that earlier untestable path no longer exists.
   • git diff 8d3767c f3df62a --stat: only claude.yml and create-porting-issue.yml change.
   • git grep -i "needs porting\|no porting" f3df62a: the only hits are the gate and its comment at create-porting-issue.yml:13,16. Nothing still names needs porting.
   • gh api repos/sillsdev/machine.py/labels/no%20porting: the label exists. needs porting returns 404, which matches the rename in the PR body.
   • gh pr view 393 --json labels: no labels. So merging this PR will file a porting issue in sillsdev/machine, which is where the PR body says the same opt-out is still needed.
5. Not verified: no YAML parse or actionlint this round. The Python parse needed an approval I did not have, and actionlint is not installed. The label trigger and the label-gated issue creation cannot be tested before merge.

Public API, optional dependency, published-wheel surface, sillsdev/machine parity: None verified. Only workflows changed. sillsdev/machine's copy of the porting gate still needs the opt-out, as the PR body says.

Findings: F1 addressed.

Reviewed at f3df62a