qinyan-paper-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external paper content, including abstracts and full text from remote URLs. This creates a surface for indirect prompt injection attacks where malicious instructions hidden within a paper could attempt to influence the agent's output. \n
- Ingestion points: The PAYLOAD variable in scripts/analyze.sh which includes paper abstracts and URLs. \n
- Boundary markers: None identified; the skill processes the full text provided by the API. \n
- Capability inventory: Execution of curl for network requests and python3 for JSON processing. \n
- Sanitization: The script performs basic JSON structure validation using python3 to prevent command injection within the JSON payload. \n- [COMMAND_EXECUTION]: The skill executes a local shell script (scripts/analyze.sh) using bash. The script takes a JSON payload as an argument and performs network requests. \n
- Evidence: SKILL.md instructs the agent to run bash scripts/analyze.sh '<json_payload>'. The script safely handles this argument by quoting it and validating the JSON format before use in the curl command. It communicates with api.qinyanai.com to perform its primary function.
Audit Metadata