The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from issue tracker titles, descriptions, and comments, which could be used as a vector for malicious instructions to influence the agent's logic and actions.
Ingestion points: SKILL.md specifies that the agent loads the map body and child tickets from the repository's issue tracker to orient itself and resolve decisions.
Boundary markers: The instructions do not define any specific delimiters or 'ignore' instructions to help the agent distinguish between the external issue content and its core operating instructions.
Capability inventory: The skill can create, update, and close issues on the tracker and may delegate tasks to other skills such as /prototype or grill-me based on the content found in the issues.
Sanitization: There is no evidence of sanitization, filtering, or validation of the data retrieved from the tracker before it is incorporated into the agent's context.

wayfinder