AgentProof
AgentProof explores independent verification of AI-produced software changes, linking requirements, test execution, evidence, defects and release decisions.
What this work covers
- Requirement-based checks and independent verification
- Evidence-backed findings and review decisions
- Public explorer illustrating the verification approach
- Demo is not a claim of complete production readiness
Scope and transparency
The public explorer is a demonstration. Individual capabilities must be evaluated against current evidence; production containment and full autonomous repair are not claimed.