Long-term memory for your AI
Don't spend tokens without keeping the knowledge
Every session with an AI assistant builds understanding — and without memory, it's gone the moment the session ends. Rohrpost remembers for you: what your AI learns today is context it starts with tomorrow. Fewer repeated explanations, cheaper calls, and answers that build on last week's work. All on your own hardware.
GET STARTED
claude plugin install rohrpostRequires Claude Code. The plugin registers the MCP server, installs skills, and configures lifecycle hooks.
curl -fsSL https://rohrpost.io/install-mcp.sh | shDownloads the rohrpost-mcp binary for your platform and adds it to your PATH. Works with any MCP-compatible client (Claude Desktop, Cursor, VS Code Copilot, Kiro).
Or configure manually
Add to your MCP client config (e.g. claude_desktop_config.json):
{
"mcpServers": {
"rohrpost": {
"command": "/path/to/rohrpost-mcp",
"env": {
"ROHRPOST_GATEWAY_URL": "https://app.rohrpost.io",
"ROHRPOST_API_KEY": "rp-your-api-key"
}
}
}
}Direct downloads:
Tokens are the only thing you pay for twice
Context you already paid an LLM to build is thrown away at the end of every session — then bought again the next morning. Rohrpost turns token spend from a running cost into an investment: knowledge in, knowledge kept.
Without memory, tokens evaporate
- Every session starts from zero — you re-explain the architecture your AI understood perfectly yesterday.
- The same files, the same constraints, the same decisions are sent again and again. You pay for the same understanding twice, ten times, every day.
- Hard-won insights — root causes, dead ends, "why we did it this way" — vanish the moment the context window closes.
With Rohrpost, tokens compound
- Every conversation is captured, distilled, and indexed — automatically, as a side effect of working.
- The next call gets a compressed brief of everything relevant instead of re-reading everything. Fewer tokens in, better answers out.
- Decisions, fixes, and constraints become a knowledge graph your whole team — and every future session — can query.
See what it looks like
Real screenshots from a running Rohrpost instance — the MCP integration your AI sees, and the knowledge graph you control.
Claude sees 11 rohrpost tools the moment it connects — ingest, recall, graph traversal, OKF export, and more. Zero custom code required.
The admin panel renders your entire knowledge graph as an interactive canvas. Drill into any entity, trace relationships, and spot gaps at a glance.
Everything you need for AI memory
Everything your AI learns, kept and made useful — under your control.
On-Prem First
Everything your AI learns stays yours. Rohrpost runs on your own server, so sensitive conversations never leave your network — and there's no per-seat cloud bill that grows with your usage.
Knowledge Graph
Your AI connects the dots. The people, files, decisions, and tools in your conversations are linked automatically — so asking about one surfaces everything related, even from a different session weeks ago.
Hybrid Retrieval
Finds it however you ask. Describe what you mean in plain language, paste an exact error message, or start from a related concept — three search strategies work together so the right context comes back every time.
MCP Native
Works with the tools you already use. One plugin gives Claude, Cursor, or VS Code a persistent memory — your assistant stores and recalls knowledge on its own, with no integration code to write.
Stream Everything
Never lose a thought. Every interaction is durably recorded the moment it happens and kept replayable — even if part of the system goes down, nothing your agents learned is lost.
Built in Rust
Light enough to run anywhere. A memory for your AI shouldn't need a data center — Rohrpost runs comfortably on a small VPS or a Raspberry Pi, with overhead you'll never notice.
Graph Explorer
See what your AI knows. Browse the accumulated knowledge visually, follow connections between topics, and spot what's there (or missing) — no query language required.
Skills Registry
Teach once, reuse everywhere. Save your best prompts, checklists, and working instructions as named skills — every agent on your team can fetch and follow them.
Open Knowledge Format
Your knowledge is portable. Export everything as human-readable documents any time — move to another instance, share with teammates, or archive for compliance. No lock-in, ever.
From conversation to context in four steps
From conversation to memory in four steps — automatically, in the background.
Capture
Every AI interaction — prompts, responses, diffs, and errors — becomes a capsule the moment it happens. The tokens you already paid for are kept, not thrown away.
Transport
Nothing gets dropped. Every capsule is delivered with guarantees and stays replayable — a complete, auditable trail of what your agents learned and when.
Distill
Raw conversation becomes knowledge. Key facts are indexed for search from every angle, and the people, files, and decisions inside are linked into a growing graph.
Recall
Next time you ask, the relevant knowledge comes back as a compact brief sized to your token budget — your AI starts warm instead of starting over.
Built for every scale
Whether you are a solo developer or an enterprise team, Rohrpost adapts to your workflow and infrastructure requirements.
Solo Developers
The Problem
You repeat the same context to your AI assistant every session. Previous conversations vanish, forcing you to re-explain architecture decisions, coding patterns, and project constraints from scratch.
The Solution
Rohrpost captures every interaction and recalls relevant context automatically. Persistent memory across sessions saves 40% on token costs by eliminating redundant context windows.
Development Teams
The Problem
Knowledge lives in individual chat histories that teammates cannot access. Onboarding new developers means re-discovering decisions that were already made and explained to an AI months ago.
The Solution
Shared knowledge graph lets any team member query the collective AI memory. The OpenAI-compatible proxy integrates without code changes — existing applications work immediately.
Enterprises
The Problem
Sensitive IP flows through third-party AI providers with no audit trail. Compliance teams cannot verify what data was shared or how AI responses influenced production decisions.
The Solution
Self-hosted deployment keeps all data on-premise with a full audit trail for compliance. Every capsule is durably stored and replayable, providing complete observability into AI interactions.
Deploy in under a minute
Start on the smallest server you have — Rohrpost is a single lightweight binary that's running before your coffee is ready. Or skip the setup entirely with the hosted version.
docker compose up