Quick facts
- Best for
- Hillock - Local Memory Engine for Offline AI Agents
- Pricing
- Free
- Editor rating
- 4.5 / 5
- Community saves
- 0
About Hillock
Hillock is an AGPL-3.0 Python project for local neuro-symbolic memory. Its README describes a three-layer architecture using a SQLite knowledge graph for triples, Hebbian associative memory, a 10,000-dimensional HDC/VSA reservoir, deterministic similarity gating, local Ollama answer generation after retrieval, and a target of running under 1.2 GB VRAM or CPU-only.
Pros
- Local neuro-symbolic memory built around SQLite knowledge triples
- Hebbian associative memory layer for co-activation patterns
- 10,000-dimensional HDC/VSA reservoir for context matching
- Deterministic similarity gate before local LLM answer generation
- Designed for CPU-only or low-VRAM offline AI setups
- Multi-hop path reasoning with HyperGraph-HDC in v0.6.0
- AGPL-licensed Python project with verification scripts
Cons
Pricing
Open-source repository
$0
- • Public source code
- • No listed software subscription for the repository
- • Users pay for their own model APIs, compute, hosting, or connected services