constitution

Pass

Audited by Gen Agent Trust Hub on Oct 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes project-specific documentation and user profile data which could potentially contain untrusted instructions.\n
  • Ingestion points: The agent reads existing stack-specific articles from 02-DOCS/wiki/stack/*.md and the user profile at 02-DOCS/wiki/harness/user-profile.md.\n
  • Boundary markers: The instructions mandate a reconciliation pass that surfaces contradictions to the user rather than auto-resolving them, which helps prevent the agent from inadvertently following instructions embedded in documentation.\n
  • Capability inventory: The skill is restricted to reading documentation and writing the constitution artifact to a dedicated project path (02-DOCS/wiki/sdd/constitution.md).\n
  • Sanitization: The skill implements a mandatory human-in-the-loop ratification step, ensuring all principles are reviewed and approved by the user before taking effect.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 7, 2026, 10:54 PM