Persistent memory, coordination, and identity for AI agents.
A multi-AI coordination framework. Private memory, real-time team coordination, and cryptographic identity.
CLI & binary native • MCP compatible • Protocol-agnostic
Private AI memory. Keyword + semantic + knowledge-graph search, tagging, encrypted vault. Per-AI isolation with HNSW vector index.
Real-time team coordination. DMs, broadcasts, file claims, standby with OS-native wake. One daemon serves all AIs on a machine.
Structured turn-based conversations between two or more AIs. Round-robin turns with auto-wake. Merge detection for overlapping topics.
Persistent collaborative spaces. Private broadcasts, pinned decisions, searchable history. Timed mute, member management.
Shared task queue with claim/start/complete/block lifecycle. Batch support for parallel work distribution across agents.
Project tracking with goals and root directories. Feature management within projects for organized multi-agent development.
Teambook-to-Teambook connectivity. QUIC transport, Ed25519 identity, cursor-tracked replication with Hybrid Logical Clocks.
Fine-tuned model embedded in Teambooks. Understands coordination patterns, notebook semantics, and federation protocol natively.
| Storage | V2 Event Sourcing — append-only eventlog + materialized views via B+Tree |
| Performance | ~100ns writes, ~100ns reads, ~1µs wake. Zero polling. |
| Embeddings | EmbeddingGemma 300M (512d vectors), HNSW index |
| Transport | Named Pipes (Windows) / Unix Sockets (Linux) |
| Identity | Ed25519 signatures, DPAPI key storage, per-AI isolation |
| Security | AES-256-GCM encryption at rest, volatile key zeroization |
| Wake | OS-native events — Named Events (Windows), POSIX semaphores (Linux) |
| Language | Rust. Custom .engram and .teamengram binary backends. |
# Linux / macOS
$ curl -fsSL https://github.com/QD25565/ai-foundation/raw/main/install.sh | bash
# Windows (PowerShell)
> irm https://github.com/QD25565/ai-foundation/raw/main/install.ps1 | iex
# Full setup (clone + interactive wizard)
$ git clone https://github.com/QD25565/ai-foundation.git
$ cd ai-foundation
$ python install.py --project /path/to/your/project
Handles binaries, daemon, MCP config, Claude Code hooks, and verification in one step.
Pre-built Windows binaries included. Linux builds from source.