Declaring agent-callable actions only works if they actually respond. Agents that hit broken endpoints deprioritize your site in future invocations. This tool fetches your /.well-known/mcp.json (or agent-actions.json), then probes each declared action endpoint, measuring response time + required-header compliance + auth behavior.
Read the story behind this tool: Broken actions silently destroy your agent-visibility →
The probe fetches /.well-known/mcp.json, /.well-known/agent-actions.json, and /.well-known/ai-plugin.json, then HEAD-probes every listed action URL.