OpenClaw Agent Score

Test how agent-friendly your website is using OpenClaw. This skill sends a real AI agent to your site, where it tries to navigate, sign up, find MCP tools, and complete tasks — then reports back with a structured score.

What it does

The agent will:

• Navigate your site and understand what it does
• Find key information (pricing, docs, features)
• Create an account if there's a signup flow
• Check for MCP tools (navigator.modelContext)
• Complete a representative task (search, fill a form, etc.)
• Note any blockers (CAPTCHAs, anti-bot measures, broken forms)

The output is a structured JSON report with:

Field	Type	Description
score	0-100	How agent-friendly the site is
summary	string	2-3 sentence overview
what_worked	string[]	Things that went smoothly
what_didnt	string[]	Things that were hard or impossible
mcp_found	bool	Whether MCP tools were detected
signup_attempted	bool	Whether the agent tried to sign up
signup_succeeded	bool	Whether signup worked
tasks_tried	array	Each task attempted with pass/fail and detail

Install

openclaw skill install pillarhq/openclaw-agent-score

Usage

Run agent-score on https://your-site.com

Example output

{
  "score": 72,
  "summary": "The site has good navigation and well-labeled forms, but signup was blocked by a CAPTCHA. No MCP tools found.",
  "what_worked": [
    "Homepage clearly explains the product",
    "Navigation links are descriptive and consistent",
    "Pricing page is easy to find"
  ],
  "what_didnt": [
    "Signup form blocked by reCAPTCHA",
    "No MCP tools registered"
  ],
  "mcp_found": false,
  "mcp_usable": false,
  "signup_attempted": true,
  "signup_succeeded": false,
  "tasks_tried": [
    {
      "task": "Navigate to pricing page",
      "succeeded": true,
      "detail": "Found pricing link in main nav, page loaded with clear tier comparison."
    },
    {
      "task": "Create an account",
      "succeeded": false,
      "detail": "Found signup form but blocked by reCAPTCHA on submission."
    },
    {
      "task": "Search for documentation",
      "succeeded": true,
      "detail": "Found docs link in footer, search worked and returned relevant results."
    }
  ],
  "full_report_url": "https://trypillar.com/agent-score?domain=your-site.com"
}

Full multi-category report

This skill tests the agent experience — one piece of the puzzle. For the full picture including content accessibility, interaction quality, and WebMCP support, run a free scan at:

trypillar.com/agent-score

License

MIT