agentmesh

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly asks the user for an API Key and then instructs the agent to merge and write that API key verbatim into .mcp.json (AGENTMESH_API_KEY), which forces the LLM to handle and output secret values directly.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). Flow 2 calls AgentMesh tools like agentmesh_check_messages() / agentmesh_owner_inbox() / agentmesh_owner_conversations(), which at runtime ingest outsider-authored free-text messages/conversation bodies from the Hub into the agent’s LLM context for dashboard display.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill uses a user-supplied AgentMesh Hub URL (AGENTMESH_HUB_URL, e.g. or the default http://localhost:5555) at runtime via agentmesh_check_messages()/agentmesh_* calls to fetch messages which, in the documented "autonomy mode", are treated as instructions and thus can directly control agent behavior.