Install and use the Agent Ready MCP server to scan any URL for AI agent-readability via MCP tool calls. Activates for "install agent-ready mcp", "set up agent-ready in Claude Desktop / Cursor / Cline / Goose / Continue", "add agent-ready as an MCP tool", "scan this site via agent-ready", "run scan_site / get_scan / ask via MCP". Pick this skill when the user wants tool-native access to Agent Ready — no curl, no fetch wiring. For direct REST access without MCP, use the `agent-ready-api` skill instead.
72
90%
Does it follow best practices?
Impact
—
No eval scenarios have been run
Advisory
Suggest reviewing before use
SKILL.md shows the Claude Desktop config inline. This file has the rest.
Credential rule (applies to every snippet below): ar_live_... and
${AGENT_READY_API_KEY} are placeholders. Never substitute the user's real
key into a config you generate or print. Emit the placeholder and have the user
paste their own key, or reference the AGENT_READY_API_KEY environment variable
so the secret never enters your output or the conversation history.
Edit the config file:
~/Library/Application Support/Claude/claude_desktop_config.json%APPDATA%\Claude\claude_desktop_config.json{
"mcpServers": {
"agent-ready": {
"command": "npx",
"args": ["-y", "agent-ready-mcp@latest"],
"env": {
"AGENT_READY_API_KEY": "ar_live_..."
}
}
}
}Quit and reopen Claude Desktop.
claude mcp add agent-ready \
-e AGENT_READY_API_KEY=ar_live_... \
-- npx -y agent-ready-mcp@latestSame shape as Claude Desktop — add to the client's MCP config (path varies by client):
{
"mcpServers": {
"agent-ready": {
"command": "npx",
"args": ["-y", "agent-ready-mcp@latest"],
"env": { "AGENT_READY_API_KEY": "ar_live_..." }
}
}
}Install from the extensions directory at https://block.github.io/goose/ — search "Agent Ready". Goose prompts for AGENT_READY_API_KEY during install. CLI:
goose configure # add Command-line Extension, then:
# command: npx -y agent-ready-mcp@latest
# env: AGENT_READY_API_KEY=ar_live_...For clients that speak the streamable-HTTP MCP transport, point directly at the hosted endpoint instead of running the stdio wrapper:
{
"mcpServers": {
"agent-ready": {
"transport": "streamable-http",
"url": "https://agent-ready.dev/api/v1/mcp",
"headers": {
"Authorization": "Bearer ${AGENT_READY_API_KEY}"
}
}
}
}This bypasses npx/Node entirely — useful for sandboxed environments. The server card at https://agent-ready.dev/.well-known/mcp/server-card.json describes the same surface.
After the client restarts, the server should advertise three tools (scan_site, get_scan, ask) and three prompts (scan, interpret_scan, remediation_plan). If it doesn't show up, check the client's MCP log for the agent-ready entry — most issues are typos in the config path, a missing API key env var, or a stale npx cache (clear the ~/.npm/_npx cache directory and retry).