The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface (Category 8) because it retrieves and processes untrusted data from the Linear platform. * Ingestion points: Data is ingested via GraphQL queries for issue titles, descriptions, comments, and document contents. * Boundary markers: No boundary markers or 'ignore' instructions are provided to help the agent distinguish between system instructions and data content. * Capability inventory: The agent can execute network requests using curl and run local Python scripts. * Sanitization: No sanitization or validation logic for the retrieved API data is specified.
[REMOTE_CODE_EXECUTION]: The instructions reference a Python helper script at scripts/linear_api.py that is not included in the provided file bundle. The skill instructs the agent to locate and execute this script from the local filesystem (~/.hermes), creating an unverifiable dependency.
[COMMAND_EXECUTION]: The skill makes extensive use of the terminal tool, instructing the agent to run curl for API operations and python3 for script execution. While necessary for functionality, these capabilities could be misused if the agent's context is manipulated by malicious data.
[EXTERNAL_DOWNLOADS]: The skill configuration in agents/openai.yaml references the official Linear MCP server at https://mcp.linear.app/mcp. This is documented as a reference to a well-known service provided by the vendor.

linear