agentleFS
sblueminsbluemin/fleet-harness/.claude/skills/console-handoff/SKILL.md

console-handoff

Skill18 starsChanged yesterdayMITView on GitHub

Prepare and seed an isolated Fleet Console and open it in this Objective's Fleet Browser so the user can try a change themselves. Use console-e2e for agent-driven browser or Electron Desktop verification.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read Build and seeding, then build changed dependencies before Console.

    line 20
  2. Distinctive:

    Create a small throwaway Theater in the scratchpad that the agent may read and edit.

    line 22
  3. Very specific:

    Use scripts/seed-console.mjs to prepare only the required state.

    line 23
  4. Distinctive:

    Read Handoff format, open the seeded URL in this Objective's Fleet Browser, leave that tab, then deliver the URL, seeded Objectives/states, …

    line 24
  5. Distinctive:

    Never stop/restart the user's Console or an unknown runtime.

    line 28
  6. Distinctive:

    Do not stop the instance later for cleanup.

    line 30

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “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

Console Handoff

Deliver a Console a person can immediately exercise: this Objective's Fleet Browser already showing the requested state, not an empty server. Unlike console-e2e, leave the handed-over instance and its Fleet Browser tab running.

Inputs

  • <worktree>: absolute checkout path whose build is being handed over.
  • <scenario>: initial state and interactions to try.
  • <model>: exact gateway model id when the scenario needs a provider. Otherwise do not pin or invoke one.

Resolve values from the request and task context. Live turns spend real quota; create only the Objectives necessary for the authorized scenario.

Procedure

  1. Read Build and seeding, then build changed dependencies before Console.
  2. Boot from an absolute binary path with a fresh runtime directory placed per console-e2e's run-directory rule. Confirm the PID command points inside <worktree> and read the lock's port with its fixed read.
  3. Create a small throwaway Theater in the scratchpad that the agent may read and edit. Never use the user's checkout or the worktree as the scenario Theater.
  4. Use scripts/seed-console.mjs to prepare only the required state. Distinguish requested state from actual seed results; adjust the prompt/fixture when they differ. Do not report a failed setup as ready.
  5. Read Handoff format, open the seeded URL in this Objective's Fleet Browser, leave that tab, then deliver the URL, seeded Objectives/states, interactions, recreation prompt, build branch/SHA, data path, model/quota use, and PID concisely. Explain first-run dialogs and Escape behavior.

Boundaries and completion

  • Never stop/restart the user's Console or an unknown runtime.
  • Open the seeded URL in this Objective's Fleet Browser. Do not drive that instance as console-e2e or open it with agent-browser. Use the OS opener only when Fleet Browser is unavailable.
  • Do not stop the instance later for cleanup. Do not close the handoff tab or the user's other tabs. The user decides when it ends. If a requested rebuild requires restart, re-read the lock, open the new URL in Fleet Browser, and send it.
  • Finish once seed state, PID, URL, and the Fleet Browser open (or the documented fallback) are verified and the handoff is delivered. Handoff is not proof that usability verification passed.

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.