cheese-brain

What This Skill Does

Cheese Brain is a high-performance, DuckDB-backed knowledge management system designed to serve as the persistent memory layer for AI agents and power users. Unlike ephemeral chat history, this skill stores structured entities—such as projects, workflows, tools, and decisions—in a searchable local database. It supports sub-millisecond keyword lookups and BM25 full-text search, allowing for nuanced relevance ranking. By centralizing disparate information into a queryable interface, it transforms scattered data into a unified, actionable knowledge graph.

Installation

To integrate Cheese Brain into your environment, use the OpenClaw hub CLI.

Command: clawhub install openclaw/skills/skills/mhugo22/cheese-brain

For manual local setup:

1. Clone the repository: git clone https://github.com/mhugo22/cheese-brain.git
2. Navigate to the directory: cd cheese-brain
3. Set up your virtual environment: python3 -m venv venv && source venv/bin/activate
4. Install the package: pip install -e .
5. Verify functionality: cheese-brain stats

Use Cases

Cheese Brain excels when you need to maintain context over long time horizons. Key scenarios include:

• Technical Documentation: Storing tool paths, API configuration details, and infrastructure credentials.
• Decision Tracking: Documenting the rationale behind past architectural choices to prevent repeating past mistakes.
• Workflow Management: Maintaining step-by-step procedures for recurring tasks to ensure consistency.
• Relationship Management: Tracking contact details, calendar feeds, and professional context.

Example Prompts

1. "Cheese Brain, what was the reason we decided to use DuckDB for the analytics module last month?"
2. "Search for the backup script workflow and provide me with the current configuration path and execution schedule."
3. "Add a new project entry for 'Alpha-Portal' with the repo URL https://github.com/org/alpha and tag it as 'priority' and 'infrastructure'."

Tips & Limitations

• Precision vs. Recall: Use the search command for broad keyword queries, but switch to fts (full-text search) when you need to rank by relevance or filter by specific categories.
• Data Structure: Utilize the --data flag with well-formatted JSON strings to ensure your metadata remains machine-readable for future programmatic access.
• Limitations: As a local database tool, it does not sync across multiple machines automatically; ensure your database file resides in a synced folder (like Dropbox or Git) if you require multi-device access. Keep your database updated regularly using the update command to ensure the agent is working with the latest context.