The Agent Skills Directory

[SAFE]: The skill consists entirely of instructional markdown files and configuration metadata. No executable scripts, persistence mechanisms, or privilege escalation attempts were found.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external pull request diffs, which constitutes an attack surface. However, the instructions are defensive in nature, explicitly guiding the agent to verify the security and correctness of the code being reviewed.
Ingestion points: Processes pull request diffs and surrounding code context as specified in SKILL.md.
Capability inventory: Uses standard agent file-reading and analysis capabilities; no dangerous tools or dynamic execution instructions are present.
Boundary markers: No specific delimiters are defined for the diff content, but the review process is manual-style reasoning.
Sanitization: The skill focuses on analytical review rather than programmatic sanitization of input.
[SECURITY_BEST_PRACTICES]: The reference materials (e.g., references/03-services-context-i-layer.md and references/04-concurrency.md) provide proactive security advice, such as using Redacted for sensitive values, limiting concurrency for user-controlled data, and enforcing runtime schema validation for all external API responses.

effect-typescript-review