The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of git diffs and source code files to provide context for review subagents.
Ingestion points: SKILL.md workflow steps 2 and 4 define the ingestion of diffs and file contents.
Boundary markers: The instructions use markdown headers like ### Git / diff output to separate untrusted data from the prompt context.
Capability inventory: The skill utilizes delegate_task to spawn subagents for analysis.
Sanitization: There is no explicit sanitization of the input content beyond the use of structural boundary markers.

thermos