COMB — Chain-Ordered Memory Base
Lossless context archival for AI agents. No more lossy summarisation.
Pitch
COMB gives AI agents lossless memory — a hash-chained, honeycomb-structured archive that never throws away context. Available on PyPI as comb-db.
Description
COMB is a honeycomb-structured, lossless context archival system for AI agents. Instead of summarising conversations (lossy), COMB archives the full text as documents in a hash-chained, three-directional graph — temporal, semantic, and social.
Key Features
- Lossless Archival: Full conversation text stored, never summarised
- Hash-Linked Chain: Cryptographic integrity across every document
- Honeycomb Graph: Temporal, semantic, and social traversal axes
- BM25 Full-Text Search: Zero external dependencies
- CLI:
comb stage,comb rollup,comb search,comb walk - PyPI:
pip install comb-db - Zero Dependencies: Pure Python 3.10+, stdlib only (core)
0 comments
No comments yet.
Sign in to be the first to comment.