Galaxy Brain

Galaxy Brain is an MCP server that blends structured thinking with sequential execution, enabling you to reason through problems, convert thoughts into actionable steps, and run those steps in order to produce results. This server is useful when you want to automate complex problem-solving workflows that involve reasoning, data transformation, and external actions.

How to use

You interact with Galaxy Brain through an MCP client by using thinking and doing workflows. Start a thinking session to reason step by step, then generate a bridge plan that converts your conclusions into executable operations. Finally, run the plan so each operation feeds its results into the next, producing a finished outcome.

Key patterns you will use include variable piping, where results from one operation feed into subsequent ones, and the ability to branch thoughts to explore alternative approaches before converging on a final plan.

A typical flow starts with a thinking session to articulate the problem, followed by creating a plan, and then executing a batch of operations that may read, transform, and write data. You can reference prior operation results in subsequent steps using a $results array.

How to install

Prerequisites: you need Python installed on your system. You can install the Galaxy Brain MCP server and then run it as a local process.

# Quick install (manual, Python-based)
pip install galaxy-brain

Then you add Galaxy Brain to your client configuration so it can be started as an MCP server.

{
  "mcpServers": {
    "galaxy_brain": {
      "command": "python",
      "args": ["-m", "galaxy_brain.server"]
    }
  }
}

Configuration and notes

Configure Galaxy Brain to control how many thoughts, operations, and branches it will allow, as well as logging levels. You can also set how results are piped and when to stop on errors.

{
  "thinking": {
    "max_thoughts": 50,
    "max_branches": 10,
    "max_revisions_per_thought": 5
  },
  "doing": {
    "max_operations": 50,
    "default_timeout": 30,
    "max_timeout": 300,
    "stop_on_error": true
  },
  "bridge": {
    "auto_execute": false,
    "validate_before_execute": true
  },
  "log_level": "INFO"
}

Environment variables let you tailor the server behavior without changing config files. Common options include log level and resource limits.

Common usage patterns include starting a thinking session, adding thoughts, revising as needed, branching to explore alternatives, and finally concluding with a plan to execute.

Security and best practices

Run Galaxy Brain in a controlled environment, restrict access to your MCP client, and avoid exposing internal operation results to untrusted users.

Review operation outputs before writing to disk or sending data to external services to prevent leaking sensitive information.

Troubleshooting

If operations fail, check the timeout and stop_on_error settings in the configuration. Review the individual operation parameters and ensure referenced results exist in the $results array.

Available tools

start_thinking

Begin a thinking session and set initial problem and parameters.

think

Add a single thought to the current thinking session with an attached confidence value.

revise

Revise a previously added thought within a thinking session and provide a rationale.

branch

Create a branch from an existing thought to explore an alternative approach.

conclude

Finish thinking and record the final conclusion with confidence.

execute_batch

Run a sequence of operations with variable piping between results.

transform_json_parse

Parse JSON content from a prior operation and produce a structured result.

json_stringify

Convert data to a JSON string for downstream use.

generate_plan

Create an execution plan from a concluded thinking session.

execute_plan

Run a generated plan to perform the actions in sequence.