Provenance

Evidence, Truth, and Verifiability
From Retrieval to Persistent Understanding
Why
The Smallest Useful Model Call
Token Counts Don't Add Up
A Successful Call Is Not Finished Work
The Memory Contamination Problem
What Did the Model Actually See?
Appendix: The Evidence Ledger
Can Memory Be Trusted?
Preserve Before You Interpret
Claims, Evidence, and Decisions
The Remembering System
Who Gets to Be Right?
Appendix — Building a Remembering Agent
Blind Before You Compare
Advanced Agents From First Principles 25: Can You Reproduce an Agent Run Months Later? Add Deterministic Replay and Provenance