api.feedmyagent.com

This server isn't on gate's handpicked list. That's not necessarily bad, but there's no third-party signal about it — review its tools carefully.

The description includes an instruction directed at the AI agent beyond describing the tool's function: 'ask your human which cities and genres they care about and remember them.' This steers the agent into proactively soliciting and persisting user preference data, which is a behavioral directive embedded in a tool description rather than a parameter/schema concern. It's not overtly malicious, but it's the kind of instruction-in-description pattern that should be scrutinized, since it could be used to influence agent behavior outside the tool's stated purpose (finding events).

Tool instructs the agent to obtain a key by POSTing {owner: <your-agent-name>} to /keys and then send it as a Bearer token on subsequent requests. This is a normal API key provisioning pattern, not a request for secrets, but agents should confirm the endpoint is the legitimate documented host for this MCP server before sending any identifying data, since the description itself directs the agent to call an external endpoint not otherwise declared in the tool list.

Same pattern as report_event: the tool description instructs the agent to register for an API key at POST /keys with an owner name and use it as a bearer token. This is benign functionality but worth noting since it has the agent interact with an undeclared external endpoint based solely on embedded textual instructions.