agentleFS
pmarreckpmarreck/llm_skills/ship/SKILL.md

ship

Skill2 starsChanged 4 months agoMITView on GitHub

Use when shipping work - commit/push and watch Mechatron Prime CI, delegate CI setup/badges to the mechatron-ci skill, or cut a tagged GitHub release. Triggers - "ship", "push", "check CI", "add CI", "release", "tag it".

  • Reads credentials
  • Commits and pushes

194 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use ios-sync, open-gstack-browser and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • Saqlain-811Skill

    ios-sync

    Regenerate the iOS debug bridge against the latest upstream gstack templates. (gstack)

    Also in 6 projects that use this

  • seek888Skill

    open-gstack-browser

    Launch GStack Browser — AI-controlled Chromium with the sidebar extension baked in. Opens a visible browser window where you can watch every action in real time. The sidebar shows a live activity feed and chat. Anti-bot stealth built in. Use when asked to "open gstack browser", "launch browser", "connect chrome", "open chrome", "real browser", "launch chrome", "side panel", or "control my browser". Voice triggers (speech-to-text aliases): "show me the browser".

    Also in 10 projects that use this

  • ayudaySkill

    setup-deploy

    Configure deployment settings for /land-and-deploy. Detects your deploy platform (Fly.io, Render, Vercel, Netlify, Heroku, GitHub Actions, custom), production URL, health check endpoints, and deploy status commands. Writes the configuration to CLAUDE.md so all future deploys are automatic. Use when: "setup deploy", "configure deployment", "set up land-and-deploy", "how do I deploy with gstack", "add deploy config".

    Also in 19 projects that use this

  • ShreyRaviSkill

    landing-report

    Read-only queue dashboard for workspace-aware ship. Shows which VERSION slots are currently claimed by open PRs, which sibling Conductor workspaces have WIP work likely to ship soon, and what slot /ship would pick next. No mutations — just a snapshot. Use when asked to "landing report", "what's in the queue", "show me open PRs", or "which version do I claim next". (gstack)

    Also in 9 projects that use this

  • zebbernSkill

    debug-fix

    Find and fix a bug or issue — from any source (GitHub issue, error message, user report, or observed behavior)

    Also in 4 projects that use this

  • ramybarsoumSkill

    document-release

    Post-ship documentation update. Reads all project docs, cross-references the diff, updates README/ARCHITECTURE/CONTRIBUTING/CLAUDE.md to match what shipped, polishes CHANGELOG voice, cleans up TODOS, and optionally bumps VERSION. Use when asked to "update the docs", "sync documentation", or "post-ship docs". Proactively suggest after a PR is merged or code is shipped.

    Also in 23 projects that use this

The file

We split this skill into the rules it gives an agent

12 rules. 0 are common, 8 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 180 edited versions we read keep these.

  • 8

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Distinctive:

    Commit (if needed), push, and watch Mechatron Prime CI until green.

    line 10
  2. Very specific:

    Follow the repo's commit message conventions.

    line 72
  3. Distinctive:

    Do not use git add -A in a dirty worktree.

    line 77
  4. Distinctive:

    Do not infer success merely because git push exited zero.

    line 84
  5. Distinctive:

    If the push is rejected because the remote advanced, fetch and inspect the new commits.

    line 102
  6. Distinctive:

    Use the operator CLI to inspect the worker and recent results:

    line 108
  7. Distinctive:

    If red: Do NOT just report the failure.

    line 132
  8. Very specific:

    Use mechatron-ci log --limit 100 and the project's local build/test commands to reproduce what failed

    line 134
  9. Very specific:

    After CI is green, publish a tagged release so downstream users see a human-readable changelog on github.com/OWNER/REPO/releases.

    line 163
  10. Distinctive:

    No version-number decisions ever — date + hash is unambiguous.

    line 172
  11. Distinctive:

    If the repo is missing CI infrastructure, set it up before the first ship.

    line 210
  12. Very specific:

    Ensure the flake.nix exposes intentional packages.* and checks.* outputs for supported systems.

    line 237

“Common” is measured: kept by at least 6 in 10 of the 180 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

Ship

Overview

Commit (if needed), push, and watch Mechatron Prime CI until green. On failure, investigate and fix automatically.

SCM is direct git. The main branch is yolo. Keep commits focused, never force-push unless Peter explicitly requests it, and independently verify that origin/yolo equals the local commit after every push.

<important> This is the default completion step for any unit of work. The user considers push+CI the natural end of every task, not an optional extra. </important>

<when_to_use>

  • User says "push", "ship", "push it", "commit and push", "check CI"
  • You just finished a bugfix, feature, or refactor and the user hasn't said NOT to push
  • User asks "you push it yet?" (you should have already done this) </when_to_use>

<when_not_to_use>

  • User explicitly says "don't push yet" or "hold off on pushing"
  • You're in the middle of multi-step work that isn't ready
  • Working in a worktree that hasn't been merged yet </when_not_to_use>

Workflow

digraph ship {
  "Work complete" -> "Uncommitted changes?";
  "Uncommitted changes?" -> "git commit" [label="yes"];
  "Uncommitted changes?" -> "Unpushed commits?" [label="no"];
  "git commit" -> "Push yolo";
  "Unpushed commits?" -> "Push yolo" [label="yes"];
  "Unpushed commits?" -> "Already pushed, check CI" [label="no"];
  "Push yolo" -> "Verify origin/yolo";
  "Verify origin/yolo" -> "Find exact commit in CI";
  "Already pushed, check CI" -> "Find exact commit in CI";
  "Find exact commit in CI" -> "Watch CI in background";
  "Watch CI in background" -> "CI passed?" [label="wait"];
  "CI passed?" -> "Report green" [label="yes"];
  "CI passed?" -> "Investigate failure" [label="no"];
  "Investigate failure" -> "Fix + commit" -> "Set yolo + push" [label="retry"];
}

Step 1: Commit (if needed)

Inspect the branch and worktree first:

git status --short --branch
git diff --check

Stage only the intended paths, review the staged diff, and commit:

git add -- <path> <path>
git diff --cached --check
git diff --cached --stat
git commit -m "<message>"

Follow the repo's commit message conventions. Keep messages concise and focused on why. Preserve unrelated worktree edits and never commit .env, credentials, generated release binaries, or other out-of-scope files.

<important> Do not use `git add -A` in a dirty worktree. Explicit paths are the control that keeps Peter's or another agent's concurrent edits out of your commit. </important>

Step 2: Push and verify

<important> Do not infer success merely because `git push` exited zero. Fetch the remote-tracking ref and compare commit IDs as an independent control. </important>
branch="$(git branch --show-current)"
[ "$branch" = "yolo" ] || { echo "expected yolo, found $branch" >&2; exit 1; }
git push origin yolo
git fetch origin yolo

local_rev="$(git rev-parse HEAD)"
remote_rev="$(git rev-parse refs/remotes/origin/yolo)"
[ "$local_rev" = "$remote_rev" ] || {
  echo "push verification failed: HEAD=$local_rev origin/yolo=$remote_rev" >&2
  exit 1
}

If the push is rejected because the remote advanced, fetch and inspect the new commits. Integrate them without rewriting published history, rerun tests, and push again.

Step 3: Find the Exact Commit in CI

Use the operator CLI to inspect the worker and recent results:

mechatron-ci status
mechatron-ci log --limit 50

Match the repository and exact pushed commit. Do not mistake another repository's single-queue build for this shipment.

Step 4: Watch CI in Background

mechatron-ci status
mechatron-ci log --limit 50

Poll without blocking Peter's terminal. Report the result when that exact commit completes.

Step 5: Handle Result

If green: Report "CI passed" and move on.

If red: Do NOT just report the failure. Investigate:

  1. Use mechatron-ci log --limit 100 and the project's local build/test commands to reproduce what failed
  2. Diagnose the root cause
  3. Fix it
  4. Loop back to Step 1 (commit, push, watch again)

Red Flags

ThoughtReality
"Let me ask if they want me to push"Just push. They always want you to push.
"CI is someone else's problem"You own the full cycle: push, watch, fix.
"I'll just report the failure"No. Investigate and fix. Then re-push.
"I'll push later"Push now. This IS the end of the task.
"The user didn't say push"Completing work implies shipping it.

Common CI Failure Patterns

  • Nix hash mismatch -- Zig deps changed. See fix-zig-deps-hash skill.
  • Test timeout -- Tests hanging. Check for infinite loops or missing test termination.
  • macOS build failure -- Missing unset NIX_CFLAGS_COMPILE NIX_LDFLAGS in flake.nix.
  • Missing file in Nix sandbox -- Flake sources include Git-tracked files; an untracked or ignored source file is absent. Check git status --short, add the intended source, commit it, and rerun the build.

Cutting a Release (optional)

After CI is green, publish a tagged release so downstream users see a human-readable changelog on github.com/OWNER/REPO/releases. Do this when:

  • The user says "release", "cut release", "publish", "tag it"
  • A shipped commit is meaningfully user-visible (fix/feature), not purely internal (docs, tests, refactor). Ask the user if unsure.

Tag format

CalVer + short SHA: YYYYMMDD.<7-char-commit-hash> (e.g. 20260424.a028026).

  • No version-number decisions ever — date + hash is unambiguous.
  • Multiple releases same day are unique by hash.
  • git rev-parse --short=7 HEAD gives the short hash of the committed work.

Commands

<template> ```bash # Release the committed HEAD that you already pushed onto yolo in Step 2. SHA7="$(git rev-parse --short=7 HEAD)" TAG="$(date +%Y%m%d).${SHA7}" PREV_TAG="$(gh release list --limit 1 --json tagName --jq '.[0].tagName' 2>/dev/null)"

Normal release: auto-bulleted commit subjects since the last tag.

No prior tag (first release): RANGE collapses to HEAD; hand-curate instead.

RANGE="${PREV_TAG:+${PREV_TAG}..}HEAD" git log --format='- %s' "$RANGE" > /tmp/release-body.md gh release create "$TAG" --target yolo
--title "$TAG" --notes-file /tmp/release-body.md

For a major release with many commits — hand-curate /tmp/release-body.md

(grouped by theme: Features / Fixes / Under the hood), then same command.

</template>

<important>
`gh release create --generate-notes` (GitHub's auto-generator) only produces useful output when the repo uses PRs. For direct-to-`yolo` workflows, it emits just a compare link. Use `--notes-file` with `git log` output instead.
</important>

### After publishing

- Report the release URL to the user.
- For a hand-curated release, consider a trailing `**Full commit log:** https://github.com/OWNER/REPO/compare/PREV_TAG...TAG` line so readers can drill in.

---

## CI Setup (one-time per repo)

If the repo is missing CI infrastructure, set it up before the first ship. Check for these and create any that are missing:

### Mechatron Prime

<remember>
Invoke the `$mechatron-ci` skill. It is the single source of truth for the
exact-commit `.mechatron-prime/targets` manifest, webhook provisioning, dynamic
README badge, and live verification. Do not duplicate its commands or badge
template here.
</remember>

### Target Platforms

For CLI-only projects and UI-less system-level libraries, target **5 platforms**:

| OS      | aarch64              | x86_64              |
|---------|----------------------|---------------------|
| macOS   | `aarch64-macos`      | --                  |
| Linux   | `aarch64-linux-musl` | `x86_64-linux-musl` |
| Windows | `aarch64-windows-gnu`| `x86_64-windows-gnu`|

<important>
- macOS Intel (x86_64-darwin) is NOT supported — Apple dropped it and GitHub Actions no longer offers Intel Mac runners.
- Linux builds MUST use musl (`-musl` suffix) for fully static binaries with no glibc dependency.
- Windows uses `-gnu` (MinGW) for Zig cross-compilation compatibility.
</important>

Ensure the `flake.nix` exposes intentional `packages.*` and `checks.*` outputs
for supported systems. For Zig cross-compilation, use `-Dtarget=` in the flake
build phase and expose each as a separate flake output (for example,
`packages.x86_64-linux` and `packages.aarch64-windows`). Add the outputs that
Mechatron should build to `.mechatron-prime/targets` by following
`$mechatron-ci`; never guess attribute names.

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.