agentsclimarketplace
Sign in

Handoff

Skill leobbaroni/cockpit/skills/handoff

End-of-phase packaging. Update PROJECT_LOG.md (changes, findings fixed/open, next steps), write how-to-run + revert guides, produce polished manuals in the requested languages (.docx or .md), and clean the repo of AI artifacts for final delivery. Use when the user asks for a change log, handout, user guide/manual, "next steps" doc, revert instructions, "log all that was changed", or delivery cleanup ("remove unnecessary/AI files"). For a promo/demo/teaser/product video of the app, use HyperFrames (/hyperframes) or Remotion (maestro's video-remotion module), not this skill. Not for code documentation like README API docs.From its SKILL.md

Install
npx -y skills add leobbaroni/cockpit --skill handoff

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

4.4 KB, as published. Nobody here has run it

Handoff

Turn a work phase into durable, human-readable deliverables. Pick the mode from the argument or infer it; when the user says "prepare everything", run log → guide → delivery in that order. (Promo/demo videos are not a handoff mode: use HyperFrames or Remotion.)

Mode: log (default)

Create or update PROJECT_LOG.md at the repo root. Append a dated section per work phase:

  • Changed — what was built/modified, by area, one line each.
  • Findings — issues discovered, each marked fixed or open.
  • Next steps — concrete, ordered; the next session should be able to start from this list alone.
  • Revert notes — for risky changes: what to undo and where (files, commits, config), so another user/agent can roll back without this conversation.

Never rewrite history in the log — append. Keep entries terse; this file is read at every session start.

Mode: guide

A HANDOUT.md (or user-named file) for the next human/agent: how to run the app (exact commands, ports, env vars by NAME only — never paste secret values), what was configured on this machine vs. in the repo, known flaky spots and their workarounds, and where the log/manual live.

Mode: manual

A polished end-user manual: how to use the app, maintain it, and troubleshoot — written for a non-technical reader, with screenshots of the real UI (capture from the live preview, not mockups).

  • Languages: produce the requested languages (e.g. EN and PT); when multiple are requested, keep the structure identical between versions.
  • Format: .docx when the deliverable is a document to send (via a docx skill if the harness has one, else generate the file directly); .md when it lives in the repo.

Promo / demo videos: HyperFrames or Remotion (the agent decides, or both)

Promo, demo, teaser, and product-film videos are NOT a handoff mode. Two animated-video toolchains may be installed; choose per the job (or combine them), don't hand-roll a bespoke pipeline:

  • HyperFrames (/hyperframes) — renders video from animated HTML/GSAP; routes to its own workflows (product-launch-video, general-video, motion-graphics, embedded-captions, etc.). Strong for bespoke UI-recreation promos, a linted check/snapshot/render loop, and template-driven product launches. Needs Node ≥ 22 and a standalone ffmpeg on PATH.
  • Remotion (maestro's video-remotion module) — React/TSX compositions rendered to MP4 via remotion render. Strong when you already have React components, want frame-precise programmatic control, or reuse an existing Remotion project; it bundles its own encoder (no external ffmpeg).

Decide from the starting asset and the look you want. When unsure, read /hyperframes (its router also covers when a specialized workflow fits) and pick from there. If neither toolchain is installed, say so and point at maestro's README dependency table rather than improvising a pipeline.

Mode: delivery

Final-delivery cleanup, in this order:

  1. List first, delete second. Enumerate candidate junk: AI-generated scratch files, mockup HTML, agent-config project files, throwaway scripts, generated media not referenced by the app. Show the list and get confirmation before deleting anything you didn't create this session.
  2. Verify .gitignore covers .env, secrets, and generated artifacts; confirm no secret values are committed anywhere in history you're about to push.
  3. Refresh README for the final structure (structure only for gitignored assets — link, don't embed).
  4. If the audience is non-technical, add a one-click launcher (e.g. a .bat/.cmd or shell script that installs deps and starts the app).
  5. Run the app once end-to-end after cleanup to prove deletion broke nothing.

Keep looking

Skills are one crate of 627,801. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.