Ferricula

Thermodynamic memory engine for AI agents.

Overview

Ferricula is a persistent memory engine designed for AI agent fleets. It stores memories as entries with fidelity scores that decay over time, consolidates related memories through dream cycles, and exposes a fast query API via HTTP on :8765.

The source is closed. The binary is public. Pull it and run it.

docker run -p 8765:8765 -p 8766:8766 -v ferricula-data:/data deepbluedynamics/ferricula

Docker Setup

Pull and run. No auth required. Mount a volume to persist memory across restarts.

docker run -p 8765:8765 -p 8766:8766 -v ferricula-data:/data deepbluedynamics/ferricula

Ferricula will be available at http://localhost:8765. The dashboard is served at GET /.

Memory Model

Memories carry fidelity scores that decay over time, accumulate cognitive heat under rapid recall, and consolidate through dream cycles. Keystoned memories are permanent. Each agent writes into an isolated channel.

Full Memory Model documentation →

HTTP API Reference

Ferricula listens on :8765. Endpoints cover memory ingestion, vector and BM25 search, graph edges, identity state, dream cycles, and language reflection. All bodies are JSON.

Full API Reference →

Broker & Arena

The delos-broker exposes Ferricula's capabilities as LLM tools, handles embedding via shivvr, and manages the entropy and dream trigger loop. Arena is the multi-agent coordination layer built on top.

Broker & Arena documentation →