light-review-rebuttal
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: Analysis of the skill instructions and scripts found no evidence of malicious patterns, data exfiltration, or privilege escalation.
- [EXTERNAL_DOWNLOADS]: The skill fetches public scholarly data from the OpenReview API (api2.openreview.net). This is a well-known service for academic peer review, and the download is handled via a standard-library Python script for legitimate calibration and research purposes.
- [COMMAND_EXECUTION]: The agent is instructed to run local Python scripts (scripts/fetch_openreview.py and scripts/rebuttal_budget.py). These scripts are part of the skill package, use only built-in Python modules, and perform restricted tasks such as character counting and API data retrieval.
- [PROMPT_INJECTION]: The skill processes external research papers as input, which represents an indirect prompt injection surface. The skill mitigates this by instructing the agent to adhere to rigorous academic rubrics (e.g., ScholarEval, NeurIPS guidelines) and to use specialized critical personas that prioritize evidence-based evaluation over potentially deceptive content in the input text.
Audit Metadata