The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data in the form of code diffs and pull requests, which could contain embedded instructions intended to manipulate the agent's output or reporting behavior.
Ingestion points: The skill ingests diff content during the code review process as described in the frontmatter and rules sections of SKILL.md.
Boundary markers: No specific delimiters or instruction-ignore guards are defined for the input diff content.
Capability inventory: The skill has the capability to create new files in llmwiki/wiki/draft/cave/ and append rows to llmwiki/wiki/index.md and llmwiki/wiki/log.md.
Sanitization: There are no instructions for sanitizing or escaping the content derived from the review process before it is written to the wiki files.

caveman-review