The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and process untrusted data from external sources, which can be manipulated to influence the agent's behavior.
Ingestion points: Step 1 in SKILL.md instructs the agent to fetch and read the full body and comments of external references (issue URLs, spec paths) provided as arguments.
Boundary markers: The instructions do not define clear delimiters or provide 'ignore embedded instructions' warnings for the data fetched from external URLs.
Capability inventory: The skill possesses the capability to write to the local file system (creating files in .efforts/) and perform network operations (publishing issues to GitHub or Linear APIs).
Sanitization: No explicit sanitization or validation of the fetched external content is described before it is used to generate ticket titles and descriptions.
[COMMAND_EXECUTION]: The skill uses tools to perform file system operations and potentially interact with CLI tools for issue tracking.
Evidence: Step 5 in SKILL.md describes writing local files and interacting with real issue trackers (GitHub, Linear), which typically involves executing API calls or CLI commands.
[DATA_EXFILTRATION]: While performing its primary function, the skill sends data derived from the codebase and conversation context to external services.
Evidence: The process involves publishing tickets to 'A real issue tracker (GitHub, Linear, …)'. Although these are well-known services, the movement of data from the local context to a remote API constitutes a data transfer pattern.

to-tickets