azure/open-regeneration-pr
Commit ai-projects regeneration output to a new branch and open a draft GitHub PR against Azure/azure-sdk-for-js main. Use when the prior five regen skills are complete and you need to push the branch and create a draft pull request via the gh CLI. Stages five logical commits (emitter output, post-emitter edits, samples, tests, changelog). Never force-pushes; always opens as draft.
npx skills add https://github.com/Azure/azure-sdk-for-js --skill open-regeneration-pr
regenerate-from-typespec, apply-post-emitter-edits, author-samples, author-tests, update-changelog) have all completed.sdk/ai/ai-projects/.| Name | Default | Description |
|------|---------|-------------|
| tspCommit | _required_ | The 40-char SHA from tsp-location.saved.yaml (the upstream commit being pulled in). |
| branchName | regen/ai-projects/<short-sha>-<yyyyMMdd> | Branch name to create. |
| remote | origin | Git remote to push to. |
gh CLI installed and authenticated against Azure/azure-sdk-for-js.sdk/ai/ai-projects/.Run from sdk/ai/ai-projects/ (the script cds to the repo root as needed).
./.github/skills/open-regeneration-pr/scripts/open-pr.ps1 -TspCommit <40-char-sha>
The script (scripts/open-pr.ps1) does:
sdk/ai/ai-projects/.regen/ai-projects/<short-sha>-<yyyyMMdd> from current HEAD.[ai-projects] regen: emitter output @ <short-sha> — generated/, tsp-location.saved.yaml.[ai-projects] regen: post-emitter edits — src/ (excluding samples/tests).[ai-projects] regen: samples for new features — samples-dev/.[ai-projects] regen: tests for new GA features — test/.[ai-projects] regen: changelog — CHANGELOG.md.origin (no force).gh pr create --draft with title [ai-projects] Regenerate from azure-rest-api-specs@<short-sha> and a body templated from changelog content + a verification checklist.https://github.com/Azure/azure-rest-api-specs/commit/<sha>.git push --force or git push --force-with-lease.sdk/ai/ai-projects/.Take azure/open-regeneration-pr from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.