The Agent Skills Directory

[SAFE]: The skill consists of text-based instructions for code simplification and contains no executable scripts, external URLs, or sensitive data access. Its functionality is restricted to providing analytical feedback on code diffs.- [INDIRECT_PROMPT_INJECTION]: The skill analyzes untrusted data in the form of code diffs. Ingestion points: Diff content provided in the agent's context. Boundary markers: Strict output format requirements (e.g., L: ) and explicit scope exclusions that protect security-related code. Capability inventory: No external tools, network access, or file-write permissions are requested. Sanitization: None. The lack of exploitable capabilities makes the risk of indirect prompt injection negligible.

ponytail-review