ldm-ai-to-human-zh
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a bundled Python script (
scripts/hanyihan_qc.py) for linguistic quality control. This script performs static analysis on text files using regular expressions to detect patterns like AI-high-frequency words, specific sentence structures, and paragraph length. The script is entirely local and does not perform any network operations or use dangerous system calls. - [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and process user-supplied text files for rewriting. This creates a surface for indirect prompt injection where instructions could be embedded in the data being processed.
- Ingestion points: Text content provided via chat or local file paths read by the agent and the Python script.
- Boundary markers: None explicitly defined in the prompt to separate user-provided text from the agent's instructions.
- Capability inventory: File system read access (via
hanyihan_qc.py) and text rewriting/modification. - Sanitization: None detected.
- As the primary purpose of the skill is text editing, this risk is considered inherent to its functionality and minimal in this context.
Audit Metadata