deep-planning-codex

Pass

Audited by Gen Agent Trust Hub on Aug 16, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill directs the agent to execute a local Python script, scripts/validate_deep_planning_artifacts.py, to verify planning artifacts. The script performs static analysis of Markdown files and does not perform network operations or access sensitive system files.\n- [PROMPT_INJECTION]: The skill processes project content and user input, which presents an indirect prompt injection surface. \n
  • Ingestion points: Project source code and repository data are read into the .deep-planning/ directory to build the evidence catalog. \n
  • Boundary markers: The workflow uses structured markdown files, explicit 'PROCEED' gates, and phase-specific status checks. \n
  • Capability inventory: The agent has access to file system operations, git commands, and the local validation script. \n
  • Sanitization: Risks are addressed through a mandatory 'Adversarial Review' phase and deterministic artifact validation that blocks execution if safety or integrity criteria are not met.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 16, 2026, 07:58 AM
Security Audit — agent-trust-hub — deep-planning-codex