GitHub Repository

Persistent memory and identity protocol for AI systems with evidence-aware recall and inspectable memory.

11 starsTypeScript

Cecil – open-source memory and identity protocol for AI

Name: Cecil – open-source memory and identity protocol for AI
Availability: InStock
Author: JohnKnopf

by JohnKnopf·Feb 27, 2026·1 point·1 comment

AI Analysis

●●SolidBig BrainZero to One

Clever multi-layer memory architecture (seed/narrative/delta), but "AI memory" is well-explored territory.

Strengths

•Three-layer identity model (seed/narrative/delta) elegantly separates immutable baseline, evolving patterns, and drift detection
•Observer agent uses light/full synthesis to compress memory without bloating context window
•Semantic retrieval via Qdrant avoids naive context stuffing that degrades reasoning

Weaknesses

•No working demo, no benchmarks comparing output quality to existing memory systems (LangChain agents, semantic-kernel)
•Unclear if the 3-LLM-call synthesis cost justifies drift detection; no evidence this outperforms simpler alternatives