BrainBox

Provides a Hebbian memory MCP server that records and recalls file usage, errors, and tool sequences to accelerate developer workflows.

typescript

6

GitHub Stars

typescript

Language

3 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "thebasedcapital-brainbox": {
      "command": "npx",
      "args": [
        "tsx",
        "node_modules/brainbox-hebbian/src/mcp.ts"
      ]
    }
  }
}

BrainBox provides a dedicated MCP server that lets you interact with a Hebbian memory model for your coding agents. It records which files you access together, tracks which errors lead to which fixes, and learns which tool sequences you use most, so it can recall relevant items instantly to speed up your workflows.

How to use

To use BrainBox as an MCP server, run the provided MCP command from your project environment. The server exposes 6 core tools for recording, recalling, error handling, and predictive capabilities, enabling your agents to fetch previously used files, fixes, and tool sequences without re-searching.

How to install

Prerequisites: ensure you have Node.js 18+ installed on macOS or Linux.

Install BrainBox via npm to fetch the MCP server tooling and hooks automatically.

Run the MCP server command shown in the integration section to start the local server.

Additional sections

Installation notes cover enabling hooks and the MCP integration in your environment. The server can be used stand-alone or integrated with Claude Code as an MCP backend. Security considerations include managing access to the MCP endpoints and keeping dependencies up to date.

Examples and troubleshooting

If you encounter issues with memory recall or missing results, verify that the MCP server is running and that your client is connected to the correct endpoint. Check the logs for tool usage and recall patterns to understand learning progress and decays.

Available tools

recall

Recall previously accessed files and related context to avoid unnecessary searches.

record

Record file interactions, edits, and tool usage to build strong synaptic connections.

stats

Show statistics about learning progress, active neurons, and decay rates.

error

Log and handle errors with contextual memory to suggest fixes.

predict_next

Predict and pre-load likely next files or actions based on recent patterns.

decay

Weaken unused connections over time to prevent memory bloat.

embed

Add vector embeddings for semantic recall of files and concepts.

hubs

Identify the most connected neurons in your network.

stale

Highlight decaying pathways and aging memory structures.

projects

Tag neurons by project to limit recall within the current scope.

sessions

List recent sessions and intents to understand usage patterns.

streaks

Track anti-recall behavior and apply decay to ignored recalls.

graph

Display an ASCII representation of the neural network.

highways

Show superhighways where frequent paths become instant recalls.

decay

Weaken unused connections over cycles to maintain efficiency.