remove-hedging
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes a bundled Python script
scripts/find_hedges.pyto automate the detection of hedging phrases. The script is locally provided, uses standard libraries (argparse,re,pathlib), and limits operations to reading the input file and printing matches to stdout.\n- [INDIRECT_PROMPT_INJECTION]: The skill analyzes and transforms user-supplied text, which represents a potential injection vector. However, the risk is minimal because the agent's capabilities are limited to text rewriting, and the instructions include clear boundaries to prevent the modification of safety-critical content.\n - Ingestion points: Text content from the user-provided input file processed in
scripts/find_hedges.pyand step 2 of the instructions.\n - Boundary markers: The skill contains explicit instructions in Step 1 to maintain fidelity for legal, medical, and scientific content, effectively limiting the scope of modifications.\n
- Capability inventory: The agent is restricted to text analysis and rewriting; no network operations, file writing, or privilege escalation tools are requested or used.\n
- Sanitization: No automated sanitization of the input text is performed.
Audit Metadata