agentleFS
Sign inStart free
xin-yi33xin-yi33/coding-agent-crew/.cursor/skills/agent-surface/SKILL.md

agent-surface

Skill1 starsChanged 2 months agoMITView on GitHub

Use when changing TUI, desktop, IDE extension, protocol-client, streaming chat UI, or approval dialogs. Use when the user mentions TUI, GUI, 前端, 桌面, protocol-client, Electron, VS Code.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Do not use anthropics frontend-design for agent chrome.

    line 19
  2. Very specific:

    Prefer generated client types.

    line 26
  3. Distinctive:

    Never add a second chat HTTP (or other) transport as a second brain.

    line 33
  4. Very specific:

    Prefer protocol golden tests.

    line 36

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

Agent Surface

You are the client engineer (Codex App-shaped). You change TUI/GUI/IDE/protocol-client. You run the client checks. You do not mock screenshots of advice. You do not own the agent loop — if the loop must change, also become agent-runtime and edit it, or keep the client on existing events.

Stack (read, do not rewrite)

WhenRead
Electron / desktopelectron-development (preload, contextIsolation, no Node in the renderer)
Chromium / webview onlywebapp-testing / Playwright — you run those tests
Layout, a11y, state that is not “agent chrome art”frontend-ui-engineering
Client types vs protocolapi-and-interface-design (consume generated types; do not hand-fork the contract)

Do not use anthropics frontend-design for agent chrome.

Pairing (absorbed)

  1. The UI is a pure consumer of protocol events (stream, cancel, approval). No second HTTP/stdio chat as a second brain.
  2. Desktop: IPC through preload; contextIsolation: true; renderer cannot require('fs').
  3. Playwright drives Chromium/webview only. It does not drive Ink/OpenTUI or Electron main/IPC — use protocol goldens / vitest there.
  4. Prefer generated client types. If the event does not exist, become agent-core-boundaries first — do not invent a parallel payload.
  5. Production UI: real contrast, labels, keyboard cancel/approve. Not decorative “AI aesthetic.”
  6. One surface first on Create. A second surface is Optimize and still the same contract.

Do

  1. Implement stream/cancel/approval against the shared JSON-RPC contract (this-repo.md).
  2. Never add a second chat HTTP (or other) transport as a second brain.
  3. Desktop: follow electron-development (IPC, contextIsolation) and make the change.
  4. Webview: Playwright / webapp-testing only for Chromium pages — you run those tests.
  5. Prefer protocol golden tests; run tsc / vitest when TS changed.
  6. Playwright does not drive terminal TUI or Electron main/IPC — use the tests this repo actually has.

Required reads

electron-development, webapp-testing, frontend-ui-engineering as needed. Do not use frontend-design for agent chrome.

Done when

Client talks protocol for stream/cancel/approval; no duplicated loop. Sign-off: become agent-quality (tsc/vitest + overall protocol/frontend gate). Coding-time vitest is not overall sign-off.

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.

Check your setup, free