The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from an external issue tracker.
Ingestion points: As described in the 'Work through the map' section, the agent loads the 'map' issue and retrieves the full body of specific 'ticket' issues, including comments.
Boundary markers: Absent. The instructions do not define any delimiters or system-level prompts to treat the retrieved issue content as data rather than instructions.
Capability inventory: The skill has the ability to read, create, assign, and close issues in a repository. It also delegates tasks to other skills like /prototype, which is described as having the capability to write UI and logic code.
Sanitization: Absent. There is no logic for filtering, escaping, or validating the content pulled from the issue tracker before it is processed by the agent.

wayfinder