The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted external data from git diffs and issue tracking systems (beads), which presents an indirect prompt injection surface. An attacker could embed instructions within a pull request diff to attempt to manipulate the review outcome.
Ingestion points: Reads code diffs via git diff and issue data via bd show/list commands in SKILL.md.
Boundary markers: The skill lacks explicit markers or instructions to treat the analyzed code diff as data rather than instructions.
Capability inventory: The skill is restricted to file system navigation, git operations, and issue tracking lookups; it lacks network access or arbitrary code execution capabilities.
Sanitization: No input sanitization or filtering is applied to the analyzed diff content.

reviewer-correctness