agentleFS
oldmilkyoldmilky/luno-for-cc/.claude/skills/browser/SKILL.md

browser

Skill0 starsChanged 4 months agoMITView on GitHub

Verify or measure the webview in a real browser. Use when the user says /browser, after any change to webview/src, or whenever a UI claim needs evidence — layout, spacing, colour, contrast, animation timing, or "does this actually render".

71 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use V3 Core Implementation, V3 DDD Architecture and 10 more.

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.
  • ruvnetSkill

    V3 Core Implementation

    Core module implementation for claude-flow v3. Implements DDD domains, clean architecture patterns, dependency injection, and modular TypeScript codebase with comprehensive testing.

    Also in 15 projects that use this

  • ruvnetSkill

    V3 DDD Architecture

    Domain-Driven Design architecture for claude-flow v3. Implements modular, bounded context architecture with clean separation of concerns and microkernel pattern.

    Also in 15 projects that use this

  • +10 more from ruvnet

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 4 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 59 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    "Should work" is not a result: If you changed something the browser renders, run this before reporting done.

    line 13
  2. Distinctive:

    Add a reply when a flow needs one the table does not cover:

    line 44
  3. Distinctive:

    Measure, do not eyeball: Read the numbers out of the page:

    line 65
  4. Distinctive:

    Read the property every frame across the transition and report what you saw — opacity at the quarter mark is what caught the exits running …

    line 84
  5. Very specific:

    Check both themes: when the change touches colour.

    line 87
  6. Very specific:

    No extension host runs. Anything whose behaviour lives on the other side of postMessage — the CLI spawn, permission decisions, MCP, …

    line 92

“Common” is measured: kept by at least 6 in 10 of the 59 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

Browser — drive the panel outside VS Code

The panel normally lives in a vscode-webview:// frame no tool can attach to. The bundle only depends on the host in two places — acquireVsCodeApi() and window.LUNO_MODE — so stubbing those boots the entire UI on localhost, where Playwright can screenshot it and read computed styles.

"Should work" is not a result. If you changed something the browser renders, run this before reporting done.

Start it

bun run harness                  # build, serve, print the URL
bun run harness --no-build       # reuse webview/dist as-is
bun run harness --mode artifact  # the plan-artifact shell instead of chat
bun run harness --port 4599

Serves http://127.0.0.1:4599/. Leave it running; rebuild and reload rather than restarting.

The seam: window.__luno

The stub is a fake host, not a silent shim. It answers the same boot handshake the extension host does, so the app reaches its chat state on its own rather than racing a timer.

CallDoes
__luno.sentevery message the webview posted to the host, in order
__luno.sentOf("prompt")just those of one type — the usual assertion
__luno.send(msg)push a host → webview message
__luno.clear()reset the recording
__luno.repliesthe fake host's reply table, editable at runtime
__luno.statewhat the app has persisted through setState
__luno.resetState()drop it, then reload for a genuinely fresh chat

Add a reply when a flow needs one the table does not cover:

window.__luno.replies.requestSkills = { type: "skills", skills: [...] };

The table the page starts with lives in webview/src/lib/harness-host.ts, typed against Inbound — a reply that no longer matches the protocol fails bun run lint instead of blanking the page at runtime. Make a lasting addition there; the runtime override above is for one-off states.

State survives a reload, in sessionStorage, the way a real webview's does. That is what makes "the chat comes back" verifiable in here — persist, reload, read __luno.state. It dies with the tab, so a fresh tab is a fresh install.

What to actually do

Prove it renders. Navigate, screenshot, look. A dangling s.someName compiles clean and renders unstyled — the screenshot is the only thing that catches it.

Measure, do not eyeball. Read the numbers out of the page:

const cs = getComputedStyle(document.documentElement);
cs.getPropertyValue("--motion-enter"); // "180ms" — as declared, not normalised
cs.getPropertyValue("--accent"); // "#d37350" on copper, the default

Drive a state you cannot reach by hand. Inject a timeline, a permission request, a plan revision — anything the host would normally send:

window.__luno.send({
  type: "timeline",
  event: { id: "1", ts: Date.now(), kind: "assistant", title: "x", body: "y" }
});

Sample an animation over time rather than trusting the duration in the CSS. Read the property every frame across the transition and report what you saw — opacity at the quarter mark is what caught the exits running on the wrong curve.

Check both themes when the change touches colour. Palettes swap through data-theme on <html>; there are seven.

What this cannot tell you

No extension host runs. Anything whose behaviour lives on the other side of postMessage — the CLI spawn, permission decisions, MCP, checkpoints, file IO — is stubbed to whatever __luno.replies says. Those need unit tests or the real Extension Development Host (F5), not this.

Clean up

Screenshots land in the repo root and .playwright-mcp/. Delete them before committing — neither belongs in git.

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.