The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from GitHub PR reviews and comments via gh pr view. This creates an indirect prompt injection surface where external feedback could influence the agent's code modifications or commit messages. Ingestion points: SKILL.md instructs the agent to poll gh pr view and gh pr checks for reviewer feedback. Boundary markers: No specific delimiters are defined to isolate untrusted PR content from the agent's internal instructions. Capability inventory: The skill has the ability to modify files, commit changes, and push to remote repositories. Sanitization: The instructions do not specify validation or sanitization of the content retrieved from GitHub before implementation.

shepherd