agentleFS
giulio333giulio333/ClaudeLens/.claude/skills/run-app/SKILL.md

run-app

Skill67 starsChanged 40 days agoMITView on GitHub

Launch and drive the ClaudeLens Electron app (mock data) via Playwright to verify UI changes or take screenshots.

13 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use release.

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.
  • MicrosoftSkill

    release

    Use ONLY when the user explicitly asks to cut/publish a Huabu desktop release or tag a version. Performs an outward action — bumps the version in apps/desktop/package.json, commits + annotated-tags vX.Y.Z, and pushes the branch + tag. That tag push auto-builds the macOS/Windows installers and publishes a GitHub Release. Provides `status`, `bump`, `tag-push`, and a one-command `run` (bump → tag-push).

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 12 edited versions we read keep these.

  • 0

    Distinctive

    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. Very specific:

    If the launch fails with spawn …/Electron ENOENT, npm skipped Electron's postinstall: node node_modules/electron/install.js.

    line 20

“Common” is measured: kept by at least 6 in 10 of the 12 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

Drive ClaudeLens either from the REPL (driver.mjs, for humans) or from a script of the same commands (run.mjs, for agents). Both share one command table in app.mjs, so a sequence worked out at the REPL pastes into a script file unchanged. Default launch uses SCREENSHOT_MODE (mock fixtures from electron/screenshotFixtures.ts — no real ~/.claude/ reads); launch real uses real data.

Setup (from repo root)

npx tsc -p tsconfig.electron.json          # build main/preload
npx vite --port 5173 &                     # renderer dev server

If the launch fails with spawn …/Electron ENOENT, npm skipped Electron's postinstall: node node_modules/electron/install.js.

Run

Interactive: node .claude/skills/run-app/driver.mjs

Scripted — the path for agents, no tty and no tmux needed:

node .claude/skills/run-app/run.mjs shots.txt      # or `-` to read stdin

Each command is awaited before the next, the run stops at the first failure (exit 1, naming the command), and quit always runs, so no Electron process is left behind. Do not pipe into driver.mjs instead: readline emits every piped line at once and the async handlers interleave.

The README gallery has a ready script — see readme-shots.txt:

SCREENSHOT_DIR=/tmp/shots node .claude/skills/run-app/run.mjs \
  .claude/skills/run-app/readme-shots.txt

Screenshots land in /tmp/claudelens-shots/ (override: SCREENSHOT_DIR). Always open and look at the screenshots — a blank frame means launch failed. Kill the Vite process when done.

Commands

commandeffect
launch [real]start app (mock data; real = real ~/.claude/)
resetreload → back to the global home
ss [name]screenshot → $SCREENSHOT_DIR/<name>.png
click-text <text>DOM-click button/link/card by text (exact, then substring)
click <css> / wait <css>DOM-click / wait for selector
text [css]print innerText (body if omitted)
eval <js>evaluate in page, print JSON
theme [light|dark]force data-theme
errorspage errors seen since launch, with stacks
sleep <ms>wait
quitclose app and exit

In a script, # comments and blank lines are skipped.

Gotchas

  • Clicks go through page.evaluate(el.click()), not coordinates — reliable with the app's custom chrome. Navigation is client-side state (no router): reach views by clicking like a user (e.g. click-text webapp → click-text Teams → click-text Open team).
  • There is no way back from a detail view — a skill, a plan and the other detail pages hide the top bar. Use reset between views instead of trying to click home.
  • Match the DOM text, not what you see. Labels like Edit, Graph, List are title-case in the markup and only uppercased by CSS, so click-text EDIT does not match.
  • theme does not disturb navigation state: navigate once, then shoot both themes from the same view.
  • The window needs ~2.5s after domcontentloaded (baked into launch); each click already waits ~1.2s.
  • An error-boundary card ("Something went wrong") carries no stack — run errors to get it. A view that only the fixtures feed can crash because the fixtures drifted behind the type, not because the view is broken.
  • On Linux add xvfb (xvfb-run -a node …); --no-sandbox is already applied.

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.