show-me-your-work
Pass
Audited by Gen Agent Trust Hub on Sep 6, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted inputs (decision text, reasons, and evidence) and writes them to a local TSV file. This surface is mitigated by input sanitization.
- Ingestion points: Arguments passed to the
scripts/log.shutility, includingdecision,why,evidence, andresultfields. - Boundary markers: The log entries are delimited using tabs in a standard TSV format.
- Capability inventory: The skill uses local shell commands to create directories and append data to files.
- Sanitization: The implementation includes a
cleanfunction that mitigates spreadsheet formula injection (CSV injection) by stripping control characters and escaping potential formula starters (=,+,-, and@) with a single quote. - [COMMAND_EXECUTION]: The skill utilizes a local shell script
scripts/log.shfor logging. The script is implemented securely, using proper variable quoting andprintfformat strings to prevent command injection from input data.
Audit Metadata