agentgauge

AgentGauge skill for Claude Code

Scan any website's AI readiness from inside your AI coding tool — then let it fix the findings in your codebase and re-scan to verify.

AgentGauge grades whether AI search can find you (citability, llms.txt, schema, AI-crawler access) and whether AI agents can use you (tools, identity, commerce signals), against the AgentReady standard with stable AR-IDs and MUST/SHOULD/MAY conformance. Free scan, instant letter grade.

Install

1. The MCP server (recommended)

claude mcp add --transport http agentgauge https://mcp.agentgauge.ai/mcp

Cursor / Claude Desktop JSON config:

{
  "mcpServers": {
    "agentgauge": { "url": "https://mcp.agentgauge.ai/mcp" }
  }
}

No auth. One tool: scan_website({ url }).

2. The skill

Copy SKILL.md into your Claude Code skills directory as agentgauge/SKILL.md:

mkdir -p ~/.claude/skills/agentgauge
curl -fsSL https://raw.githubusercontent.com/GlassCaseOfEmotion/agentgauge-skill/main/SKILL.md \
  -o ~/.claude/skills/agentgauge/SKILL.md

Use

/agentgauge scan mysite.com

or just ask: "check my site's AI readiness and fix what's failing". The skill scans, prioritizes the findings (spec MUSTs first), implements fixes in your repo, and re-scans after deploy to verify the grade actually moved.

What gets checked

31 checks across 6 categories: Discoverability, Content for Agents, AI Visibility (GEO), Capabilities (MCP/A2A/OpenAPI), Identity & Access, and Commerce.

---

Built by AgentGauge · Weekly monitoring: Watch by AgentGauge · Contact: support@agentgauge.ai