agentleFS
vchelaruvchelaru/Gum/.claude/skills/gum-release/SKILL.md

gum-release

Skill659 starsChanged 2 months agoMITView on GitHub

Gum release/build orchestration — drives the in-repo release checklist and hands off the notes draft. Triggers: cutting a release, publishing the Gum NuGet packages, releasing the tool, /gum-release.

The file

Files
SKILL.md

Invoking product-manager agent to drive the Gum release checklist.

Gum Release Orchestrator

This skill is a guided checklist driver, not full automation. Most release steps are inherently human — taking screenshots, clicking GitHub Actions, uploading to FTP, announcing on Discord/Twitter/Bluesky. The skill's job is to walk the maintainer through the steps in order, track what's done, and invoke the one step a skill can actually do (the notes draft).

Source of truth — do not restate it here

The release process lives in docs/contributing/building-and-releasing-gum.md (published to the Gum GitBook space). That doc is canonical and maintainer-edited. Read it at the start of every run and follow whatever it says today — do not hardcode or paraphrase its steps into this skill, or the two will drift.

How to run

  1. Read the doc above and confirm the current step list with the user.
  2. Ask if this is a hotfix — an emergency tool release for a bad bug, outside the normal cadence. If yes, follow the doc's "Hotfixes" subsection: skip gum-monthly-release (write a short manual note instead), skip screenshots and community announcements, and confirm with the user that the fix carries no breaking changes before tagging.
  3. Track progress — create one task per step (TaskCreate) so nothing is dropped across the long-running release, and mark each completed as the user confirms it. For a hotfix, only create tasks for the steps that apply.
  4. Walk the steps in order, one at a time. For each, state what the user needs to do (the doc has the detail) and wait for confirmation before advancing.
  5. Hand off the automatable step (non-hotfix only). Release-notes generation is the only step with real automation, and it lives in gum-monthly-release. That skill is disable-model-invocation: true, so the Skill tool refuses it — read .claude/skills/gum-monthly-release/SKILL.md and follow it yourself instead. Everything else (NuGet workflow trigger, Build-and-Release workflow, screenshots, migration doc, announcements, FRB FTP upload) is the user's manual action; you confirm and check it off.

Gotchas

  • The two processes in the doc — NuGet publishing and tool release — are independent. Ask which the user is doing; don't assume both.
  • A hotfix is a mode of the tool release, not a separate process — same steps, minus notes/promotion. Don't run gum-monthly-release for one.
  • gum-monthly-release only writes the notes markdown. It does not bump versions, cut the tag, or trigger workflows — those are separate manual steps in the doc.
  • Never offer or start the public release build, or a monthly NuGet publish, until the notes are approved: every open question answered and the migration page written or confirmed unneeded. The test build may run earlier. Create one task per step with TaskCreate and follow the doc's order; "keep going" from the user is not approval to skip a step.
  • The Full Changelog compare link in the notes can only be filled after the tag exists, so it stays a placeholder until the release is cut.

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.