The Agent Skills Directory

[SAFE]: The skill provides instructions for reviewing code changes against project standards. No malicious patterns, such as command injection, data exfiltration, or obfuscation, were found in the provided file.
[PROMPT_INJECTION]: The skill is designed to process untrusted data (pull requests and code diffs), which constitutes a potential surface for indirect prompt injection.
Ingestion points: Pull requests and TypeScript/JavaScript code changes provided by users.
Boundary markers: None explicitly defined in the instructions to separate code data from instructions.
Capability inventory: The skill is permitted to use Bash, Grep, Read, and Glob tools.
Sanitization: No explicit sanitization or filtering logic is mentioned for the ingested data.

typescript-review