qinyan-nature-polishing
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run a local Python script (
scripts/style_audit.py) to evaluate the writing style of the manuscript. The script's source code is provided within the skill package and only performs non-destructive regex-based analysis of the text. - [DATA_EXPOSURE]: The included audit script reads content from a file path provided as an argument. While designed to process academic manuscripts, it could potentially be used to read other files if the agent is misdirected, although the script only outputs style-related metadata and findings.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-provided academic text. It lacks explicit delimiters or instructions to ignore embedded commands within the input text, which is a standard attack surface for indirect prompt injection.
- Ingestion points: User-provided academic passages (processed for translation/polishing).
- Boundary markers: Absent; the instructions do not specify how to isolate user text from agent instructions.
- Capability inventory: Local command execution (
python scripts/style_audit.py). - Sanitization: Absent; the skill performs direct linguistic transformation on the input text.
Audit Metadata