Hippocampus provides human-like memory for digital data. It uses events with linked memories, each of which have their own significance rating on an open scale, reinforced through recall, and a sleep cycle to shed memories that are no longer worth keeping. It is a lossy data store by design, meant for long-term storage.
🔭 See it running — hippocampus-demo.com
A store that forgets is only worth having if what it keeps is what you turn out to need. That is measurable, so it has been measured — by replaying an agent workload fitted to a real corpus into a live instance and scoring the survivors against the standard cache-replacement baselines at the same store size:
Every access-based policy is statistically indistinguishable from random at retaining the memories that matter but are not touched often. LRU scores 20.2% against random's 19.9%; LFU manages 18.4%. Hippocampus scores 27.6% at the same store size, and +11.1 points over LRU at a larger one.
Importance is not in the access log, so a policy reading only the access log cannot see it. Method, baselines, the checks that stop it being circular, and the limitations: Retention quality.
The guides below cover the service itself; the components that ship around it are in the second table. If you are deciding rather than deploying, read Use cases & deployment modes first — its Worth knowing before you start section is the set of properties that shape what you can build on this.
| Guide | Description |
|---|---|
| 🎬 Getting Started | Step-by-step build, initial config, and first gRPC/HTTP requests. |
| 📐 Use Cases & Patterns | Embedded vs. centralised topologies and data transfer strategies. |
| 🧠 Memory Consolidation | Deep dive on decay algorithms, capacity targets, and summarisation. |
| 🧙 Configuration wizard | Build a config and its deployment artefacts in the browser, with a live forgetting preview. |
| ⚙️ Configurability | Exhaustive key reference for TLS, auth, storage drivers, and listeners. |
| 🛠️ Operations & Deployment | Containers, Kubernetes, packages, sizing, backups, shutdown, and observability. |
| 🔒 Security | What is off by default, auth and role tiers, hardening checklist, and where content can leave. |
| 📈 Retention quality | What a bounded store keeps, measured against the standard cache-replacement baselines. |
| 📊 Performance Benchmarks | Throughput sweeps across SQLite, Postgres, and MySQL under heavy loads. |
| 🧪 Demonstrations | The hosted demo, plus worked scenarios using real-world data shapes and generators. |
| 🖥️ Web console | The console every instance serves at /ui — what each tab answers, and where its numbers come from. |
| Guide | Description |
|---|---|
| 🤖 MCP Server | Give an LLM host (e.g. Claude, ChatGPT, Gemini, etc) memory tools via the Model Context Protocol. |
| 🔌 Event Sourcing | Bridge NATS, MQTT, RabbitMQ, Kafka, or the Bluesky firehose in, storing each message as a memory. |
| 🚦 Ingestor | Stage data at the edge and promote completed events into a central store under CEL rules. |
| 💻 CLI | Drive a running service from the shell over either transport. |
| 🧬 Clients & Codegen | Generate a Python, TypeScript, or any-language client from the proto or OpenAPI document. |
| 📓 Obsidian Integration | Use Hippocampus as a memory layer for an Obsidian vault via the plugin or the MCP bridge. |