takomi-codex

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill invokes several local PowerShell scripts (scripts/takomi-detect.ps1, scripts/takomi-policy.ps1, scripts/takomi-board.ps1, scripts/takomi-pi-dispatch.ps1, scripts/takomi-harness.ps1) to perform environment detection, policy loading, and task orchestration.
  • [DATA_EXPOSURE]: The skill reads configuration and requirements from project directories (.pi/, docs/) and user-specific paths (~/.pi/agent/, ~/.agents/skills/). It explicitly defines that global/user paths are read-only unless the user provides explicit write approval.
  • [PROMPT_INJECTION]: The skill processes project-level documentation and task packets to drive subagent behavior, representing an indirect prompt injection surface.
  • Ingestion points: Reads docs/project_requirements.md, .pi/settings.json, and markdown roadbooks.
  • Boundary markers: Not explicitly defined for ingested content.
  • Capability inventory: Can execute local PowerShell scripts and write task artifacts to the project directory.
  • Sanitization: Not specified, but the skill requires explicit user consent for non-diagnostic commands and file modifications.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 09:02 AM
Security Audit — agent-trust-hub — takomi-codex