resume

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill executes shell commands (git checkout, cat) and a local Python script (setup_experiment.py) using variable interpolation for experiment paths ({domain}, {name}). If these variables are derived from untrusted input without validation, it could lead to command injection.\n- [INDIRECT_PROMPT_INJECTION]: The skill reads content from local project files that could potentially be influenced by external data or previous automated steps, creating a surface for indirect instructions to be processed by the agent.\n
  • Ingestion points: .autoresearch/{domain}/{name}/config.cfg, .autoresearch/{domain}/{name}/program.md, and .autoresearch/{domain}/{name}/results.tsv (SKILL.md).\n
  • Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present in the logic when reading these files.\n
  • Capability inventory: The skill possesses capabilities for Git branch manipulation, file system reads, and local script execution (SKILL.md).\n
  • Sanitization: No evidence of content sanitization or structured schema validation is provided for the data read from the experiment files.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 10:00 PM