rigor-reviewer
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill follows secure patterns by explicitly restricting its own capabilities to a narrow set of tools (Read, Write, Glob, Grep) and prohibiting network operations or code execution. The instructions establish a clear, restricted operational boundary within a local directory path provided by the user.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted research artifacts from a directory for analysis. This ingestion represents a surface for indirect prompt injection, but the potential for malicious impact is mitigated by the skill's structured extraction logic and the absence of tools capable of exfiltration or system modification.
- Ingestion points: Files located in the user-specified artifact directory (e.g., PAPER.md, logic/claims.md).
- Boundary markers: The agent is instructed to parse specific semantic fields and record its findings in a structured JSON report.
- Capability inventory: File system access via Read, Write, Glob, and Grep tools.
- Sanitization: The skill translates the unstructured text of the research artifact into a structured JSON review without executing any embedded code.
Audit Metadata