rw-research-learning
Pass
Audited by Gen Agent Trust Hub on Jul 21, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements strict exclusion policies for sensitive files such as credentials, private keys, and environment variables. The Python script 'research_learning_scan.py' filters these by filename and extension.
- [COMMAND_EXECUTION]: Local command execution is limited to the provided analysis script and does not allow for arbitrary shell commands or privilege escalation.
- [SAFE]: Data exfiltration is mitigated by design; the skill workflow emphasizes local processing and the script contains no networking functionality.
- [PROMPT_INJECTION]: The tool processes untrusted data from user documents (ingestion in 'research_learning_scan.py') which could contain instructions intended to influence the agent's research capability assessment. While it lacks explicit prompt delimiters or content sanitization, the restricted local capability keeps the risk profile minimal.
Audit Metadata