braindb

What This Skill Does

BrainDB is a sophisticated, persistent, and semantic memory layer specifically engineered for the OpenClaw AI ecosystem. Unlike standard session-based memory that clears upon resets or compaction, BrainDB provides long-term recall by leveraging a high-performance vector database. It uses a 768-dimension semantic search architecture that identifies conceptually relevant information rather than relying on simple keyword matching. By categorizing information into episodic, semantic, procedural, and associative types, it creates a structured knowledge graph that evolves alongside your workflow. With a 98% accuracy rate and a sub-20ms latency, it serves as an always-on cognitive enhancement for your AI agent.

Installation

To install, execute the command: openclaw plugin install braindb. Alternatively, you can clone the repository directly into ~/.openclaw/plugins/braindb and execute the provided install.sh. The system requires Docker and approximately 4 GB of available RAM. During the initial installation, the installer will back up your existing configurations, initialize an isolated Docker network containing Neo4j and the embedding service, and patch your openclaw.json to integrate the plugin automatically. The process typically takes 3–5 minutes for the initial model download, with subsequent reloads taking roughly 10 seconds.

Use Cases

1. Long-term Project Context: Maintain a persistent history of project requirements, technical debt, and coding preferences across multiple development sessions.
2. Personal Knowledge Base: Offload facts, API credentials, or specific configuration nuances that the AI needs to recall weeks or months later.
3. Procedural Skill Retention: Save step-by-step instructions or custom workflows learned during a session, allowing the AI to execute complex tasks later without needing to re-read documentation.

Example Prompts

1. "BrainDB, recall the architecture pattern we decided on last Tuesday for the login service."
2. "What are the specific constraints I mentioned regarding the database schema for the new module?"
3. "Summarize everything you know about the project roadmap we've been building over the last three weeks."

Tips & Limitations

• Tips: Utilize the optional --scan command before performing a bulk migration of workspace files to preview exactly what data will be indexed. Ensure you have sufficient Docker resources allocated, as the embedding engine is memory-intensive.
• Limitations: BrainDB is strictly local; it is not designed to sync across multiple machines. Avoid storing sensitive high-clearance credentials directly in raw text; while the system is secure and localhost-isolated, utilize environment variables where possible. The system requires occasional maintenance to clear stale entries if the database grows excessively large.