37signals-rails-style
Pass
Audited by Gen Agent Trust Hub on Mar 6, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill defines a set of coding conventions for the agent to apply when reviewing or generating code, which creates a standard surface for indirect prompt injection common to coding assistants.\n
- Ingestion points: User-provided Rails source code and architectural requests handled by the agent using these style guidelines.\n
- Boundary markers: The skill does not define explicit delimiters to separate user input from the agent's instructions.\n
- Capability inventory: The agent is tasked with generating and refactoring code based on the provided patterns, including controllers, models, and authentication logic.\n
- Sanitization: No sanitization or validation logic for the input source code is provided in the skill instructions.\n- [EXTERNAL_DOWNLOADS]: The skill metadata includes an installation command that retrieves its instruction file from an external repository.\n
- URL:
https://raw.githubusercontent.com/majiayu000/claude-skill-registry/main/skills/data/37signals-rails-style/SKILL.md\n - Method: The skill file is downloaded using
curland placed in the local skills directory.
Audit Metadata