poke-integrations
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides the mechanism to connect an agent to various external platforms (GitHub, Notion, Gmail, etc.) and custom MCP servers. Data retrieved from these services is inherently untrusted and could contain malicious instructions.
- Ingestion points: Built-in integrations (Linear, Notion, GitHub, Gmail, etc.) and custom MCP URLs (e.g.,
https://example.com/mcp). - Boundary markers: The instructions do not define specific delimiters or instructions to treat external data as untrusted.
- Capability inventory: The skill enables management of integrations which grant the agent read and write access to the user's external accounts via the Poke platform.
- Sanitization: No data sanitization or validation logic is specified for data retrieved from these integrations.
- [COMMAND_EXECUTION]: The skill documents the use of the
pokeCLI for adding custom MCP servers (poke mcp add). While this is a documented platform feature for extending agent capabilities, it involves the execution of commands that process remote URLs and user-supplied API keys.
Audit Metadata