agentleFS
oldmilkyoldmilky/luno-for-cc/.claude/skills/ship/SKILL.md

ship

Skill0 starsChanged 4 months agoMITView on GitHub

Full delivery pipeline for non-trivial work — plan, implement, gates, browser evidence, independent review, report. Use when the user says /ship or asks to take a feature end to end. Never commits.

  • Reads credentials

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

6 rules. 0 are common, 4 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 181 edited versions we read keep these.

  • 4

    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:

    Read CLAUDE.md. If the task touches webview/src/, read docs/TOKENS.md too.

    line 20
  2. Distinctive:

    Before writing code, state:

    line 30
  3. Distinctive:

    No colour, radius, shadow or duration literals — tokens

    line 50
  4. Distinctive:

    Do not run prettier by hand.

    line 61
  5. Very specific:

    If the work touched src/core/* or src/services/*, it is unit-testable and should have a test.

    line 76
  6. Distinctive:

    Never invent a test that passes regardless.

    line 79

“Common” is measured: kept by at least 6 in 10 of the 181 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

Shipping end to end. Task: $ARGUMENTS

This is the full ceremony: plan → implement → gates → browser evidence → independent review → report. No auto-commit; the user reviews the diff and runs /gitpush.

Each phase has a gate. On a failure, decide between fix-and-retry (cheap and obvious) and stop-and-ask (risky or ambiguous). When in doubt, ask — one question is cheaper than 400 lines in the wrong place.

Phase 0 — Pre-flight

  1. Read CLAUDE.md. If the task touches webview/src/**, read docs/TOKENS.md too.
  2. git status --porcelain — start from a known state. Uncommitted work from something else means stop and ask, because the final diff will not be yours.
  3. Nothing long-running needs to be up. Unlike a web app, this project has no dev server to depend on: /browser starts its own harness in Phase 5. If the user happens to be running bun run dev:webview, leave it alone.

Phase 1 — Plan

Before writing code, state:

  • Which files get created or modified, with paths
  • Which side of the postMessage boundary the work lives on — extension host, webview, or both. If both, name the messages that cross
  • Which existing primitives, tokens and motion presets get reused instead of new ones
  • What will prove it works: which test, and what the browser harness will show

Gate 1. Trivial work — a single-file fix, a copy change — proceeds without approval. Anything touching multiple files, the protocol, or the permission path gets the plan shown as a todo list and stops for confirmation.

Phase 2 — Implement

Host first when both sides are involved: the protocol shape it exposes is what the webview then consumes.

The rules that matter here, all from CLAUDE.md:

  • No colour, radius, shadow or duration literals — tokens
  • Motion is a spread preset from design/motion.ts, never a hand-written transition, even when the number would match
  • Icons come from design/icons.tsx; tooltips from design/primitives/Tooltip
  • New protocol messages are added to the union in webview/src/lib/rpc.ts and handled host-side — an unhandled message fails silently, which is the worst kind
  • src/core/* imports zero VS Code APIs. Keep it that way; that is what makes it testable

The PostToolUse hooks format the file and check the comment policy as you write. Do not run prettier by hand.

Phase 3 — Gates

bun run lint    # tsc over BOTH projects, then eslint, then stylelint
bun run test    # the count in CLAUDE.md is the floor

Gate 3. Failures in files you wrote: fix, up to three attempts. Failures in files you did not touch: stop and report — do not "fix" them inside a feature change. New eslint warnings are worth a line in the report, not a stop.

Phase 4 — Tests for the change

If the work touched src/core/* or src/services/*, it is unit-testable and should have a test. If it touched only rendering, it usually is not — say so.

Never invent a test that passes regardless. A missing test named in the report is better than a placeholder that always goes green.

Phase 5 — Browser evidence

Skip only if nothing under webview/src/** changed — and say so.

Otherwise run /browser: start the harness, drive the state the change affects, and come back with something measured. A screenshot proves it rendered; a computed value proves it is right.

Gate 5. A dangling class name, a dead keyframe or a killed opacity all look correct in the diff and wrong on screen. This phase is the only place they surface.

Phase 6 — Console sanity

In the same harness run, read the console. A missing favicon is noise. A React key warning, an act() warning, a failed asset, or a thrown error is yours until proven otherwise.

Phase 7 — Independent review

Dispatch the reviewers that match what you touched, in parallel, briefed with the file list:

  • src/** → extension-security-reviewer
  • webview/src/** → webview-design-reviewer

They run in their own context and read the diff cold, which is the point — you have been staring at this for an hour.

Gate 7. CRITICAL or HIGH: stop, do not write a "ready to commit" report, fix and re-run from Phase 3. MEDIUM/LOW: proceed, but quote them verbatim in the report so the user sees them before committing.

Skip only for work with no surface at all — a doc edit, a comment. Say so in one line. When unsure, run it; it costs a minute.

Report

## /ship — <feature>

Phases:  plan ✅ · implement ✅ · gates ✅ · tests ✅ · browser ✅ · review ✅

### Changed
- [path](path) — why

### Crossing the boundary (if any)
- `<message>` — webview → host, handled in `panel.ts:<line>`

### Evidence
- gates: lint clean, <n> passed / <n> skipped
- browser: <what was driven, and the number that was measured>
- review: <verdict per agent>

### Concerns raised, not blocking
- <verbatim from the reviewers>

### Left undone
- <anything skipped, and why>

### Next
Review the diff, then `/gitpush`.

Hard rules

  • Never commit or push. That is /gitpush, and it is the user's call.
  • Never skip a phase silently. Skipping is fine; skipping quietly is not.
  • Never touch .env, credentials, or ghost.one/.
  • Never invent a test to fill Phase 4.
  • Stop on ambiguity.

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.