agentleFS
ahmadrosidahmadrosid/nakama/.agents/skills/verify-bs/SKILL.md

verify-bs

Skill826 starsChanged 12 days agoMITView on GitHub

Verify the current UI change in BrowserOS (browseros-neo MCP) on a dev app this session starts from its own worktree, on its own ports, with the user's ~/.nakama data. Use when the user says /verify-bs, "verify using browseros", or asks to see a web change working in the real app.

  • Reads credentials

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    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. Distinctive:

    Do not use or stop the user's own app on 3003 / 4310.

    line 15
  2. Very specific:

    Find what to check: Read the diff (git diff main...HEAD plus uncommitted changes).

    line 19
  3. Very specific:

    Get a worktree on the code to check: Use this session's worktree if it has one.

    line 22
  4. Very specific:

    Prefer test records you create (name them verify-bs …) and clean them up at the end.

    line 61
  5. Distinctive:

    Stop your servers: when the check is done, unless the user wants them kept: stop the two background tasks you started.

    line 65

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Verify in BrowserOS on this session's own dev app

Each session runs its own API and Vite from its own worktree, so several sessions can check changes at the same time. All of them use the user's ~/.nakama data. Do not use agent:ui, Playwright or the built-in browser. Do not use or stop the user's own app on 3003 / 4310.

Steps

  1. Find what to check. Read the diff (git diff main...HEAD plus uncommitted changes). Name the screen, the route, and the 1–3 states the change touches. Find the route in apps/web/src/App.tsx.
  2. Get a worktree on the code to check. Use this session's worktree if it has one. Else git worktree add ../nakama-verify-<branch> <branch>. A branch can be checked out in one worktree only; if another has it, use git worktree add --detach <path> <branch>. New worktree: run bun install in it first. Never switch the branch of a checkout that is not yours.
  3. Pick two free ports. Web from 3010 up, API from 4311 up. A port is free only when both are empty: lsof -nP -iTCP:<port> -sTCP:LISTEN and curl -s -o /dev/null -w '%{http_code}' --max-time 2 http://localhost:<port>/ (prints 000).
  4. Start the API in the background from the worktree root. Do not use bun run dev:server: it kills whatever listens on 4310, which is the user's API.

    set -a; . ~/.config/nakama/dev-secrets.env; set +a
    NAKAMA_CONFIG_DIR=$HOME/.nakama NAKAMA_PORT=<api> bun run apps/server/src/index.ts

    Ready when the log says Nakama server listening on http://127.0.0.1:<api>.

  5. Start Vite in the background from apps/web:

    NAKAMA_SERVER_URL=http://127.0.0.1:<api> bun x vite --port <web> --strictPort

    Ready when http://localhost:<web>/ returns 200. A "Port is already in use" error: pick the next port and start again.

  6. Open BrowserOS. Load the browseros-neo tools. name_session (category testing-and-qa), tabs action="list", then tabs action="new" with http://localhost:<web>/<route>. Never work in the user's own tabs. Cookies on localhost are shared across ports, so the user's login carries over.
  7. Drive it. snapshot mode="interactive" → act by ref → screenshot for each state. Wait with wait for="text", not a fixed pause.
  8. Do what the check needs. Save, create, edit, send or delete when the goal needs it; no need to ask first. The app uses the user's real ~/.nakama data, so:
    • Prefer test records you create (name them verify-bs …) and clean them up at the end.
    • Report every write you made to the user's existing data.
    • Close any tab the app opened in your group (for example an OAuth page).
  9. Stop your servers when the check is done, unless the user wants them kept: stop the two background tasks you started. Every API on ~/.nakama also manages the same background workers (channels, automations) and writes ~/.nakama/runtime/server-url.txt, so do not leave one running.
  10. Report. Numbered list: each state checked, pass or fail, one line each. Give the URL you used. List the data you changed. Name what you did not check and why. Keep your BrowserOS tab open.

Notes

  • Built-in plugin changes (packages/plugins/*) show only after the plugin is reinstalled on your dev app.
  • The service worker does not run under Vite dev, so the "Notify me when a reply is ready" switch stays disabled there.
  • A bug you find: say what you saw and where; fix it only if the user asks.

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.