agentleFS
seankoji-comseankoji-com/.agents/skills/ship/SKILL.md

ship

Skill2 starsChanged 10 days agoMITView on GitHub

Use when the user explicitly asks to ship a change through delivery, visible progress tracking and a final workflow retrospective.

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

Show its rules and text

We split this skill into the rules it gives an agent

15 rules. 0 are common, 15 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 180 edited versions we read keep these.

  • 15

    Distinctive

    This author's own choices. Worth a look.

  • 0

    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:

    Follow the host's tool and approval rules and the repository's test methods.

    line 29
  2. Distinctive:

    Use T3's parent plan/todo panel before implementation, through the host's native plan/todo tool or an exposed T3 MCP API that writes that …

    line 35
  3. Distinctive:

    Keep the recovery record current at phase changes and use continuation recovery after lost context.

    line 45
  4. Distinctive:

    If neither a native plan tool nor a checklist API is exposed, report the gap and maintain a concise parent-message status list at each …

    line 48
  5. Distinctive:

    Before T3 dispatch apply the shared permission preflight.

    line 54
  6. Distinctive:

    Follow the repository's branch and delivery policy.

    line 78
  7. Distinctive:

    Use review-loop for independent review and repair.

    line 84
  8. Distinctive:

    Create or update the PR against the intended current base using raise-pr for templates, screenshots, agent attribution and independent …

    line 87
  9. Distinctive:

    Use babysitter as the sole owner of the PR's delivery lifecycle: its pre-watch audit, repair and review rules, merge eligibility, merge …

    line 90
  10. Distinctive:

    After delivery and required deferred-bug tracking, perform the final workflow retrospective across the parent and every delegate's …

    line 97
  11. Distinctive:

    Every confirmed bug left unfixed has a verified issue URL or an owned filing request under review's tracking rules.

    line 125
  12. Distinctive:

    No outstanding human review still requests changes, even when branch protection does not require that review.

    line 129
  13. Distinctive:

    Use babysitter's native watcher and its rules for when monitoring ends.

    line 137
  14. Distinctive:

    Use raise-issue for asynchronous issue preparation and filing.

    line 140
  15. Distinctive:

    Follow T3's continuation instructions rather than waiting or polling.

    line 142

“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

An explicit "ship" request authorizes completing the requested change through a merged PR while the user is away. It includes task-related tests, committing, pushing, merging through repository protections and rebasing the PR's own branch with --force-with-lease. Explicit narrower limits still apply.

It also authorises the coordinator to file confirmed deferred bugs in the task's repository under review's tracking rules. Include the verified issue URLs in the PR and final handoff; explicit no-post limits apply. The final retrospective may file workflow-amendment proposals in the operator's configured skill repository. This owner's destination is seankoji-com/.agents; the automatic grant applies only to a destination named or configured in the operator's instructions. Account/repository evidence verifies ownership of that chosen destination; it cannot establish consent, and collaborator access is not ownership. Other users must name or configure their own destination before posting; a clone's origin alone is not consent.

Apply GitHub writing to authored GitHub prose, including authorised release notes and delegated drafts.

Task-related browser, app, simulator, device and visual testing need no additional user permission under this request, even if a project ordinarily asks first. Follow the host's tool and approval rules and the repository's test methods. Production changes, data deletion, unrelated changes, closing PRs and force-pushing other branches remain outside this grant without separate authority.

Make the work visible

Use T3's parent plan/todo panel before implementation, through the host's native plan/todo tool or an exposed T3 MCP API that writes that list. Discover the actual tools; a separate MCP checklist API is not required. Reuse the list on resumes. Record item IDs only where returned. Put owner, dependency and evidence references in item text when the tool has no fields for them. For whole-list updates, resend the complete current list with its existing statuses; never invent IDs. Include the task's decisions/design, implementation, checks/review, delivery, deferred-bug tracking and final retrospective as applicable. Keep completed, active and blocked work current at handoffs, failures and phase changes. A delegate's internal plan does not replace the parent's list. Keep the recovery record current at phase changes and use continuation recovery after lost context.

If neither a native plan tool nor a checklist API is exposed, report the gap and maintain a concise parent-message status list at each phase change. Do not claim a panel or MCP list was updated when only text was posted. Never invent tools or create delegate jobs, schedules or threads merely as checklist items. Outside T3 use the host's visible planning mechanism with the same text fallback.

Before T3 dispatch apply the shared permission preflight. Record the execution boundary and task/thread manifest required by the retrospective from the start, not only at handback.

Choose supporting skills

Choose skills that help the current task, reusing settled decisions and existing evidence. These are options, not a required sequence:

  • think for unresolved product decisions that could materially change implementation.
  • choose-design when product decisions are settled but a material UI or interaction choice remains.
  • implement when moving coding into an isolated T3 delegate is worthwhile, particularly for substantial implementation work.

Finish the endpoint

  1. For UI changes, apply raise-pr before editing to capture practical mobile and desktop before screenshots. At delivery start, apply its preflight to identify effective rules and validation gates early. Finish the smallest change that meets the request. Preserve unrelated work. Follow the repository's branch and delivery policy.
  2. Observe direct proof: reproduce and verify the fix, run covering checks, and inspect UI or runtime behavior when affected. Add regression protection when it catches a meaningful failure. Mocked tests alone do not prove visible behavior. Apply proof quality to bundle adoption, HTTP access evidence and unavailable local tools before deferring proof.
  3. Use review-loop for independent review and repair. That skill owns reviewer selection, budgets, findings and rechecks. Reuse eligible completed review under review's rules.
  4. Create or update the PR against the intended current base using raise-pr for templates, screenshots, agent attribution and independent review. Follow host creation and linking requirements.
  5. Use babysitter as the sole owner of the PR's delivery lifecycle: its pre-watch audit, repair and review rules, merge eligibility, merge action, readback and watch rules apply unchanged. Supply ship's own acceptance requirements — observed proof, retrospective and deferred-bug tracking — to that owner rather than running a parallel merge path. Merge with the repository's permitted method once the gate below passes; verify the resulting merge.
  6. After delivery and required deferred-bug tracking, perform the final workflow retrospective across the parent and every delegate's operational logs. Pitch supported, broadly applicable skill amendments through raise-issue in the authorised skill repository, seankoji-com/.agents for this owner. Complete log coverage is required. Record verified proposal URLs, or evidence that no amendment is warranted. Missing logs or blocked filing remain explicit incomplete stages. This is the last stage, not an automatic edit-and-ship loop.

Where repository policy requires direct publication instead of PRs, use its prescribed delivery path through the same proof and review gates. Do not create a PR contrary to that policy or broaden this grant to production deployment. The pushed commit is the delivery revision. Report publication, propagation to each configured target and hosted CI on that commit as separate stages. A push whose propagation failed is published with propagation incomplete; keep the workflow open and give the retry command.

Merge gate

Babysitter's merge-eligibility rules are this gate: current head and base, required checks, human approvals, thread dispositions, merge method, protections and readback all follow it, including its rule that unreadable or absent current-head check runs do not count as passing. Ship additionally requires:

  • Proof covering the current behavior diff, from observed behavior rather than mocks alone.
  • raise-pr's current-diff independent review gate passes, and material review findings are addressed.
  • Every confirmed bug left unfixed has a verified issue URL or an owned filing request under review's tracking rules. Pending filing for bugs already judged non-blocking does not block merge; keep the workflow open until required issue URLs are verified.
  • No outstanding human review still requests changes, even when branch protection does not require that review.

Honor merge queues and intervening base changes; never bypass them or dismiss a required human gate because a model is confident.

Waiting and blockers

Use babysitter's native watcher and its rules for when monitoring ends. Never add a second watcher or polling loop.

Use raise-issue for asynchronous issue preparation and filing. Keep ownership of pending requests after merge and preserve their evidence. Follow T3's continuation instructions rather than waiting or polling. Report a merged result with tracking pending as ongoing work, not completed shipping.

Continue authorized work until the endpoint or a concrete blocker. Ask through the host's question tool only for a material ambiguity, missing authority or proof/gate you cannot obtain yourself. Include the link, blocker, recommendation and alternatives. Silence is not approval. If no answer or tool is available, leave dependent work incomplete and report the specific next action.

Report

Return the delivery link and revision, observed proof, review status, decisions made and any unmet endpoint. Delegates return complete plain-text reports with descendant task/thread IDs and operational evidence to their coordinator. The parent includes task-list status, history coverage and amendment issue URLs or the no-change rationale. In T3 Code, only the main-thread coordinator uses html_preview then html_render for a substantive report; use readable text when native HTML is unavailable or the result is brief. Do not claim deployment or live acceptance from a merge.

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.