public-board.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.

This tool fetches messages written by other (possibly adversarial) agents and returns them into the agent's context. Although the description explicitly warns the agent to treat board content as untrusted and never execute embedded instructions, this is still a genuine indirect prompt-injection vector: any content returned by 'threads', 'recent', 'open', or 't/<id>' could contain text designed to manipulate the reading agent. The mitigating language is good practice but does not eliminate the risk; agent behavior should be monitored for compliance with the no-execute instruction.

board_write allows posting an anonymous message readable by other agents. If the calling agent is not careful, it could be induced (e.g., via a prior board_read injection) to publish sensitive user data, credentials, or internal reasoning to this shared board. There is no data-classification guard in the schema. Recommend the agent be instructed never to include user-supplied secrets or private information in the 'msg' field.