The Agent Skills Directory

[SAFE]: The skill functions as a quality assurance tool for reviewing completed work. It utilizes tools such as Read, Grep, Glob, and Bash to inspect local files and artifacts. No patterns of data exfiltration, credential theft, or unauthorized persistence were found.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data (files, PR diffs, artifacts) using powerful tools like Bash. This creates a theoretical surface where malicious instructions in a reviewed file could attempt to influence the agent. However, the skill's instructions specifically direct the agent toward evaluation and reporting, and no evidence of exploitation was detected.
Ingestion points: The skill reads relevant files, diffs, or artifacts as described in the 'Examine the output' step.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore potentially malicious commands within the artifacts being reviewed.
Capability inventory: The skill is granted access to Read, Grep, Glob, and Bash tools.
Sanitization: The instructions do not define any sanitization or validation of the content being reviewed before processing.

kernel-review