agent-reproduce-align

Pass

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/run_pair_capture.sh executes user-provided shell commands using bash -lc. This is the core functionality of the skill for capturing traces of different agents.
  • [COMMAND_EXECUTION]: The skill invokes scripts located in a neighboring skill folder (agent-reproduce-feature) to capture HTTP traces and agent state. These are standard inter-skill dependencies within this repository.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The scripts/normalize_trace.py script processes HTTP traces but explicitly redacts and drops sensitive information such as authorization headers, cookies, and provider request IDs before outputting the normalized results.
  • [SAFE]: All operations, including file reading, script execution, and trace comparison, are performed locally within the project's working directory or temporary directories as configured by the user.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 12, 2026, 08:38 AM
Security Audit — agent-trust-hub — agent-reproduce-align