The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from an external issue tracker to guide agent actions and decision-making.
Ingestion points: Retrieves content from a "map" issue and individual "decision tickets" stored on the project's issue tracker (e.g., GitHub Issues) to determine next steps.
Boundary markers: While the skill uses Markdown structure (e.g., ## Question, ## Destination) to organize content, it lacks explicit defensive prompts to ignore instructions that may be embedded within the user-provided ticket descriptions.
Capability inventory: The skill can invoke several tools including /pstack-swarm for delegated task execution, /domain-modeling for codebase edits, and /research for data gathering.
Sanitization: No validation or sanitization is performed on the data fetched from the issue tracker before it is processed by the model.
[EXTERNAL_DOWNLOADS]: The skill documentation in UPSTREAM.md references the mattpocock/skills repository on GitHub as its primary source and adaptation base.

wayfinder