The Agent Skills Directory

[SAFE]: The skill's instructions focus on technical analysis and reporting of code changes. No malicious intent or suspicious patterns were found.
[DATA_EXFILTRATION]: The skill reads local file content (git diffs and source snippets) to provide context for reviews. This is constrained to the tool's intended use and does not involve sending data to unknown external destinations.
[PROMPT_INJECTION]: The skill ingests untrusted code from git repositories. While this presents an indirect prompt injection surface, the workflow is limited to analytical reporting and does not include instructions that would lead to the execution of processed content.

review-delta