paper-fraud-auditor
Pass
Audited by Gen Agent Trust Hub on Jun 7, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION]: Paper data is transmitted to the MinerU service (mineru.net) for processing. This communication is a documented core feature and is necessary for the skill's operation.\n- [EXTERNAL_DOWNLOADS]: Installation is performed by fetching the skill package from the vendor's GitHub repository.\n- [PROMPT_INJECTION]: The skill processes external scientific papers, which presents an attack surface for indirect prompt injection via embedded malicious instructions.\n
- Ingestion points: Conversion results from MinerU are ingested from the local file system into the agent context (SKILL.md).\n
- Boundary markers: The audit logic uses a structured evidence ledger, but does not use explicit delimiters around the raw text content of the paper.\n
- Capability inventory: The agent can execute forensic scripts, write JSON and PNG evidence files, and communicate with the MinerU API.\n
- Sanitization: Numeric forensics scripts rely on regular expression matching for numbers, which limits the execution of non-numeric injection payloads during analysis.
Audit Metadata