agentleFS
Sign inStart free
FabienGreardFabienGreard/TouchGrassBar/.agents/skills/release/SKILL.md

release

Skill4 starsChanged 54 days agoMITView on GitHub

Release TouchGrassBar to production when the user requests a patch, minor, or major release. Complete backend deployment, signed builds, publication, and update-feed checks.

We split this skill into the rules it gives an agent

5 rules. 0 are common, 3 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 183 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use scripts/release.ts as the single source of truth for release checks and actions.

    line 8
  2. Distinctive:

    Use the level already given in the conversation.

    line 14
  3. Distinctive:

    Keep validation, signing, protected environments, and immutable tags in effect.

    line 22
  4. Distinctive:

    Read the release level from the conversation.

    line 29
  5. Very specific:

    Run bun run release LEVEL --execute.

    line 32

“Common” is measured: kept by at least 6 in 10 of the 183 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Release

Use scripts/release.ts as the single source of truth for release checks and actions.

Authorization

A release request and one explicit level authorize this complete production workflow. Ask for patch, minor, or major only when the level is absent. Use the level already given in the conversation.

This authorization includes release preparation, required fixes and fixtures, commit and push to main, the production backend deploy and canaries, tag creation, workflow dispatch, environment reviews by the authenticated permitted reviewer, and publication. Carry it through retries and required candidate fixes. Execute these steps without further confirmation prompts.

Keep validation, signing, protected environments, and immutable tags in effect. Resolve routine failures within this scope. Request user action only for a concrete external blocker, such as missing credentials, an unknown target, a review that the current identity cannot submit, or a change outside release scope.

Workflow

  1. Read the release level from the conversation. Completion: one level is explicit.
  2. Prepare a clean main candidate. Preserve unrelated work. Follow the fixture lifecycle and Release-note: rules in docs/release.md. Commit and push the candidate, wait for successful CI for that exact commit, and run bun run release LEVEL. Correct incomplete or technical notes before continuing. Completion: all preflight checks pass and the preview reports the current tag, next tag, commit, CI run, and accurate user-facing notes.
  3. Invoke convex-deploy-guard, identify and announce the production target, and run a dry deployment from the clean candidate checkout. Deploy that exact backend, then verify one successful production synchronization and one populated production read. This applies to every release because the desktop and backend share synchronization validators. Completion: record the deployment, candidate commit, deploy result, synchronization result, and read result before tag creation.
  4. Run bun run release LEVEL --execute. Completion: the script reports that the immutable tag was pushed.
  5. Find the tag run with gh run list --workflow release.yml --branch TAG --limit 1. Verify the tag, commit, successful CI, and pending environment before submitting its permitted review through gh api. Continue watching the build. Completion: the exact tag workflow succeeds and creates the signed, notarized draft. If a tag fails, keep it immutable and prepare the next valid version under the same release authorization.
  6. Inspect the draft with gh release view TAG --json url,isDraft,body,assets. Verify the expected DMG, updater archive and signature, checksums, database compatibility evidence, and release trust evidence. Check that the description matches the changes, names the correct DMG, and includes collapsed technical verification and the full changelog. Correct the description when needed. Completion: the draft and its assets pass all release checks.
  7. Complete the configured publication workflow, including any permitted public-release review. If publication is manual, publish the verified draft with gh release edit TAG --draft=false --latest. Preserve its tag, title, assets, and prerelease state. Fetch the public release and stable latest.json endpoint again. Completion: the release is public, the update feed names this version and its signed archive, and the final report gives the release URL and verified result.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free