No-Proof No-Action Gate
A public evidence boundary for high-risk AI action claims: when required proof is missing, the system should refuse action instead of implying a recommendation.
Concept Index
Stable names for public evidence.
These entries bind recurring terms to public pages, DOI records, GitHub repositories, Hugging Face artifacts, and explicit evidence boundaries.
Robot/Vision Downgrade Rule
A public boundary for robot and vision evidence: a clean recognition frame is not deployment proof; uncertain perception should downgrade to review or stop.
Self-Certification Grounding Check
A public grounding check for AI self-certification claims: self-reporting must land in records, failures, review status, and expiry conditions.
Proof-Carrying Action
A technical entry point for proof-carrying action: action claims should carry evidence, authority, boundary, and no-go records.
WisdomBench
A public benchmark and evidence line for longitudinal failure-learning, grader drift, and wisdom-oriented evaluation boundaries.
The Grounding Canon
A public evidence-ladder entry for claim-to-evidence reporting, grounding boundaries, and current P46/P43/P44/P45 links.