gpt-engineer
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill architecture is built around local orchestration, task planning, and telemetry analysis, with no evidence of unauthorized external communication.
- [SAFE]: Implements a proactive security guard in assets/codex/hooks/gpt_engineer_guard.py to intercept and block high-risk shell commands such as 'git reset --hard' and 'git push --force'.
- [SAFE]: Features a robust data sanitization mechanism in scripts/run_journal.py that identifies and redacts API keys, tokens, and passwords from persistent logs using a comprehensive regular expression policy.
- [SAFE]: Command execution is appropriately scoped to platform-provided binaries (codex) and standard version control tools required for the skill's primary engineering functions.
- [SAFE]: The bootstrap process in scripts/bootstrap.py manages local configurations and agent profiles using atomic writes and lifecycle locking to maintain system integrity while preserving user-customized files.
- [SAFE]: The skill uses structured handoffs (handoff.schema.json) to ensure consistent and verifiable data exchange between parent and child agents, reducing the risk of schema confusion.
Audit Metadata