The Agent Skills Directory

[SAFE]: The skill provides structured guidelines for naming analysis and does not contain any malicious logic, obfuscation, or unauthorized access patterns.
[EXTERNAL_DOWNLOADS]: The README mentions installation via npx skills add ratacat/ratacats-skills, which targets a repository belonging to the author ('ratacat') and represents standard functionality for this platform.
[PROMPT_INJECTION]: The skill is designed to ingest and analyze user-provided artifacts such as codebases and plans, which constitutes a surface for indirect prompt injection. However, the instructions mandate a review-only role and conceptual boundary mapping, which reduces the risk of the agent executing malicious instructions embedded in the analyzed data. 1. Ingestion points: User-provided codebase files, plans, architecture docs, and tickets. 2. Boundary markers: The skill instructs the agent to identify conceptual boundaries but lacks explicit instructions to use delimiters or ignore embedded instructions in the analyzed data. 3. Capability inventory: No file-writing or network tools are authorized in the configuration; the skill is restricted to providing analysis and recommendations. 4. Sanitization: No specific input sanitization methods are mentioned in the skill instructions.

name-review