CIPS Stack Complete

The whole cognitive architecture. One install.

CIPS Stack is CASCADE + PyTorch Memory + Hebbian Mind + CMM unified. Four systems that work together as one cognitive

infrastructure. Memory that persists, searches, learns, and synthesizes.

The Problem: Building AI memory from scratch means choosing between dozens of databases, vector stores, and caching

layers. They don't talk to each other. You spend months on plumbing instead of building.

The Solution: A complete cognitive architecture designed to work together from day one.

The Stack:

• CASCADE - 6-layer memory with temporal decay (sub-1ms)

• PyTorch Memory - GPU vector search (sub-2ms)

• Hebbian Mind - Self-organizing neural graph

• CMM - Unified search across all backends (one query, all systems)

• Soul Matrix integration - Pre-token gating via Hebbian relationships

What Makes It Different:

• Unified search - CMM queries all backends, synthesizes results

• Dual-write patterns - save once, persist everywhere

• Temporal decay + importance scoring - memories fade unless reinforced

• Hebbian learning - the system learns your patterns without training

• Docker deployment - one command, full stack running

Performance (real benchmarks):

• 847 memories, 179 ops/sec

• 1.23ms ping, 3.51ms median search

• Sub-millisecond CASCADE reads

• Sub-2ms PyTorch semantic search

What's Included:

• All four components, full source

• Docker Compose for one-command deployment

• MCP servers for Claude Desktop

• Integration examples

• Architecture documentation

• 1 year updates

Requirements: Python 3.10+, Node.js 18+, Docker, CUDA GPU (optional, CPU fallback available)