poke-integrations

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly allows adding arbitrary MCP-compatible servers via poke.com/integrations/new (see references/custom-mcp.md), and Poke calls those third‑party MCP servers to expose tools/resources/prompts that the agent will read and act on, so untrusted external content can influence agent behavior.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill allows adding a custom MCP server (e.g. https://example.com/mcp) which Poke "calls ... like any other MCP client" at runtime and the MCP exposes tools/resources/prompts that can directly control agent instructions, so the fetched remote content can influence/drive the agent's behavior.