agentleFS
IgorGanapolskyIgorGanapolsky/mac-yolo-safeguards/.agents/skills/chrome-webmcp/SKILL.md

webmcp-agent-readiness

Skill3 starsChanged 4 days agoMITView on GitHub

Build, audit, and test browser-native WebMCP tools with Chrome document.modelContext or declarative form annotations; use for Chrome WebMCP, agent-friendly web actions, journey evaluation, browser tool schemas, or modelContext integration.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    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. Very specific:

    Use WebMCP as progressive enhancement: the human interface and ordinary DOM automation must still work when document.modelContext is …

    line 8
  2. Very specific:

    Do not call a browser or site WebMCP-ready from its version or source code alone.

    line 21
  3. Distinctive:

    Test deterministic validation and a real Chromium implementation:

    line 40
  4. Distinctive:

    Add app-specific unit, browser, and evaluation cases.

    line 48
  5. Distinctive:

    Read the tool and security contract before exposing tools cross-origin, auto-submitting forms, or registering write/external-action tools.

    line 50
  6. Distinctive:

    After changing this skill, mirror the tested copy across agent homes:

    line 54

≈ “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

Chrome WebMCP

Use WebMCP as progressive enhancement: the human interface and ordinary DOM automation must still work when document.modelContext is unavailable.

Start with a real capability receipt

python3 ~/.agents/skills/chrome-webmcp/scripts/probe_webmcp.py --browser all

Current measured routing on this Mac:

  • Chrome Stable 151 and Canary 154: imperative registration, discovery, and execution pass in isolated headless profiles with WebMCP test flags.
  • BrowserOS Neo 148: document.modelContext is unavailable; keep using its DOM/CDP path until a fresh probe passes.

Do not call a browser or site WebMCP-ready from its version or source code alone. Require the probe or an equivalent page-level registerTool → getTools → executeTool receipt.

Build workflow

  1. Choose one non-overlapping function per tool. Prefer a declarative annotated form for an existing form; use the imperative API for app logic, navigation, diagnostics, or stateful actions.
  2. Copy assets/webmcp-kit.js for imperative tools or adapt assets/declarative-form.example.html for forms. Feature-detect the API and preserve the ordinary UI fallback.
  3. Describe the intended tools and journey evaluations in a manifest based on assets/manifest.example.json. Static-only authoring checks are explicit:
python3 ~/.agents/skills/chrome-webmcp/scripts/lint_webmcp_manifest.py path/to/webmcp-tools.json --static-only
  1. Produce a fresh isolated browser receipt, then run the readiness audit. A manifest without that receipt is unverified and exits nonzero; static validity alone is never ready.
python3 ~/.agents/skills/chrome-webmcp/scripts/probe_webmcp.py --browser all --output /tmp/webmcp-probe.json
python3 ~/.agents/skills/chrome-webmcp/scripts/lint_webmcp_manifest.py path/to/webmcp-tools.json --probe /tmp/webmcp-probe.json
  1. Test deterministic validation and a real Chromium implementation:
python3 ~/.agents/skills/chrome-webmcp/scripts/test_manifest_lint.py
node ~/.agents/skills/chrome-webmcp/scripts/test_webmcp_kit.mjs
python3 ~/.agents/skills/chrome-webmcp/scripts/probe_webmcp.py --browser chrome
  1. Add app-specific unit, browser, and evaluation cases. Prove UI state changes separately from the tool return value.

Read the tool and security contract before exposing tools cross-origin, auto-submitting forms, or registering write/external-action tools.

System install

After changing this skill, mirror the tested copy across agent homes:

bash ~/.agents/skills/chrome-webmcp/test.sh
bash ~/.agents/skills/chrome-webmcp/scripts/sync_agent_homes.sh

Relationship to sibling capabilities

  • This skill is the fleet-wide readiness gate and runtime proof source.
  • webmcp-instrument-site is a Claude-side authoring reference; route its readiness claims through this skill's linter and probe.
  • agent-mystery-shopper audits a customer journey and stops before unauthorized external submission. It does not prove WebMCP support.
  • The daily funnel-truth LaunchAgent proves URL/status/body availability only, not tool discovery, execution, or journey completion.

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.