External cognition governance harness for coding agents: claim verification, memory continuity, risk routing, bounded improvement, and adapter contracts. Current: v0.20.2. DOI: 10.5281/zenodo.21189879.

Audit-first static console for passive HEA/MPEA candidate review, descriptor-level risk checks, and claim-boundary-safe report snapshots.

Reflexive case study of artifact inflation, epistemic grounding failure, and claim-boundary control in non-expert AI-assisted research.

Claim-boundary audit for code-review feedback utility claims.

Auditable DPO evaluation framework for LLM-as-judge reliability, held-out validation, and claim-boundary reporting.

Claim-bounded reduced-surrogate archive for Socio-Traffic Thermodynamics: stop-and-go waves, synchronization risk, and risk-budgeted information release in toy traffic networks. No policy or deployment claim.

Public restart surface for VOT / The Pleasure Order.

Claim-bounded failure-boundary audit of an A10-derived Hubble-tension scaffold: no physical E(z)/H(z) bridge, no likelihood comparison, and no solution claim.

Public showcase for a fail-closed molecule identity, metadata consistency, and claim-boundary audit framework.