cloudflare_temp_email/.claude/skills/cf-temp-mail-release/SKILL.md

name, description

name	description
cf-temp-mail-release	Create a GitHub release for cloudflare_temp_email project. Use when the user asks to create a release, publish a version, tag a release, or make a new release. Reads CHANGELOG.md for release content, collects merged PRs via `gh` CLI, and creates a properly formatted GitHub release.

Release Workflow

Steps

1. Read version: Get current version from worker/package.json ("version" field) and the latest release tag via gh release list --limit 1.
2. Read CHANGELOG: Read CHANGELOG.md for the current version section (e.g. ## v1.4.0(main)). Verify content matches CHANGELOG_EN.md. If entries are missing from either file, notify the user.
3. Collect PRs: Get the last release tag timestamp, then filter merged PRs by time:

   TAG="$(gh release list --limit 1 --json tagName --jq '.[0].tagName')"
   SINCE="$(git show -s --format=%cI "$TAG")"
   gh pr list --state merged --search "is:pr is:merged merged:>$SINCE base:main" --json number,title,author --limit 200
   

   Sort by PR number ascending.
4. Check database upgrades: Compare the previous release tag with the exact release target, including db/ migrations, db/schema.sql, and the Admin database-update implementation. Use git diff <previous-tag> <target> -- db/ (fetch missing refs first), not PR timestamps alone. Read each changed migration to determine required upgrade steps and effects on existing data. Only recommend the Admin database-update button after verifying it applies these migrations. Do not execute database changes as part of release preparation.
5. Compose release body: Follow the template in references/release-template.md. Key rules:
   • Write release body in bilingual format: Chinese sections first (from CHANGELOG.md), then wrap the English sections (from CHANGELOG_EN.md) in <details><summary>English</summary>...</details>.
   • Copy changelog sections verbatim from both files. Omit empty sections.
   • For features requiring database migrations, follow release v1.5.0: add exactly three concise, natural-language sub-bullets under each corresponding feature in both languages. In order, explain how to enable it; the release-tag-pinned SQL migration link or verified Admin page action, plus whether unused features can skip it; and what the feature does. Do not prefix these lines with labels such as "Feature switch:", "Database changes:" or "Functionality:". Keep each feature's three lines separate, without a shared Database Changes section or classifying optional migrations as Breaking Changes. Preserve the original changelog bullet; these sub-bullets supplement it.
   • State which feature requires each migration. Verify the feature's disabled code paths before saying it is optional: if the feature is not used and existing functionality does not depend on the migration, explicitly say users can skip it unless they enable that feature later. Distinguish these optional migrations from migrations required for normal operation; do not describe every migration as mandatory for all deployments.
   • Keep migration instructions accurate and short: never recommend reapplying an already-applied non-idempotent migration. Include any destructive data impact in the database sub-bullet rather than hiding it to meet the format. Resolve unclear upgrade paths before publishing; do not invent switches or migration actions.
   • Wrap PRs list in <details><summary>PRs</summary>...</details>.
   • Always include the cache-clearing discussion link.
   • End with **Full Changelog** comparison link.
6. Create release:
   • Write body to a temp file (e.g. /tmp/release-notes.md)
   • Run: gh release create vX.Y.Z --title "vX.Y.Z" --notes-file /tmp/release-notes.md --target main
7. Verify: Read back the published body, verify all required migration links and bilingual upgrade instructions are present, then confirm the release URL and ask the user to review.