Your OpenClaw Agents Finally Have a Memory Worth Keeping
Brainfork brings sovereign, durable memory to OpenClaw agents — so they remember your work, your decisions, and your context without you repeating yourself every session.
Phil Bennett
Head Brainforker
If you've been running OpenClaw agents for a while, you'll have hit the wall. The agent does great work — but tomorrow it wakes up fresh, with no idea what happened yesterday. You've been copy-pasting context into prompts, maintaining a MEMORY.md by hand, watching files drift out of sync. It works, until it doesn't.
That's exactly the problem Brainfork was built to solve.
What Brainfork Actually Does
Brainfork is a hosted memory layer for AI agents. It stores your documents, decisions, and retrieval context in one place — a server you control — and makes that memory available to your agent at the start of every run.
For OpenClaw specifically, there's now a first-class plugin that slots directly into the memory system. Three lines of config and your agent stops starting from scratch.
1plugins:2 slots:3 memory: brainfork-openclaw4 entries:5 brainfork-openclaw:6 config:7 baseUrl: https://api.brainfork.is8 endpoint: your-server-endpoint9 apiKey: bf_live_...10 autoRecall: true11 autoIndex: true12 captureDecisions: trueWith autoRecall on, Brainfork injects relevant context before each agent run — not by dumping your whole archive into the prompt, but by retrieving what's actually relevant. With autoIndex and captureDecisions enabled, your MEMORY.md and daily notes sync automatically, and important decisions get logged as they happen.
The Sovereignty Bit (This Part Matters)
Here's the thing that separates Brainfork from just bolting a vector database onto your setup: the knowledge stays yours.
When you use most memory solutions, your context ends up in someone else's infrastructure with no clean way to inspect, export, or migrate it. Brainfork gives you a server you control. The documents are portable, the decisions are preserved, and none of it is tied to a single model vendor. Switch models, switch agents — your memory follows.
That's not a small thing. Agents that know your work are genuinely useful. Agents that know your work but live in a black box you can't audit are a risk.
What Gets Better With This
Once you have Brainfork wired up to OpenClaw, a few things just start working the way you'd hoped they would:
Context without prompt engineering. Instead of writing elaborate system prompts to remind your agent what project it's working on, that context gets recalled automatically from your indexed documents and previous decisions.
Decision continuity. When your agent makes a significant call — choosing a framework, setting an architecture direction, deciding how something should behave — that gets logged. Next session, it knows why things are the way they are.
MEMORY.md that actually stays current. Right now, keeping MEMORY.md meaningful requires manual curation. With autoIndex running, changes sync automatically with deterministic hashing, so there's no drift between what's on disk and what's in the memory layer.
One memory provider, not a sidecar. Brainfork handles retrieval over MCP, so OpenClaw doesn't need a separate parallel API. Clean, and one less thing to break.
Getting Set Up
- Create a Brainfork server at brainfork.is — free trial, no card required for 14 days.
- Grab your API key and endpoint from the server management page.
- Install the plugin:
openclaw plugins install @brainfork/brainfork-openclaw - Drop the config into your OpenClaw config file (see above).
- Let it run. Your agent will start recalling context on the next session.
The plugin handles the rest — pushing documents, logging decisions, and pulling relevant context before each run.
Pricing
The base plan is €5/month and includes 25k requests and 1,000 indexed documents per month. Overage is €1 per 10k extra requests and €1 per 250 extra indexed documents. For most individual setups and small teams, the base plan will comfortably cover you.
If you're running OpenClaw agents and you've been fighting with memory drift, this is the fix. Your agents do better work when they remember what you've already figured out. Now they can.
Ready to own your AI knowledge?
Start building your personal AI infrastructure today.
Get Started Free