Galaxy Brain
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| GALAXY_BRAIN_LOG_LEVEL | No | The log level for Galaxy Brain | |
| GALAXY_BRAIN_MAX_THOUGHTS | No | Maximum number of thoughts allowed | |
| GALAXY_BRAIN_MAX_OPERATIONS | No | Maximum number of operations allowed |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| start_thinkingA | Start a new thinking session to reason through a problem step by step. Returns a session_id to use with other thinking tools. |
| thinkA | Add a thought to a thinking session. Use this to reason through a problem step by step. Set next_thought_needed=false when done. |
| reviseA | Revise a previous thought in a thinking session. Use when you realize an earlier thought was wrong or incomplete. |
| branchA | Create a new branch to explore an alternative approach. Useful when you want to try a different direction without losing your current train of thought. |
| concludeA | Conclude a thinking session with a final synthesis. Call this when you've finished reasoning through the problem. |
| execute_batchA | Execute a batch of operations sequentially with variable piping. Use $results[N].path.to.value to reference previous results. Available services: python, shell, file, transform. |
| execute_singleB | Execute a single operation (convenience wrapper for execute_batch) |
| list_servicesB | List available services and their methods for execute_batch |
| generate_planC | Generate an action plan from a completed thinking session. Converts thoughts into executable operations. |
| execute_planC | Execute a generated action plan |
| think_and_doC | Complete cognitive loop in one call: think through a problem, then execute operations. The ultimate galaxy brain move. |
| get_sessionC | Get the current state of a thinking session |
| list_sessionsC | List all thinking sessions |
| get_planC | Get a plan by ID |
| list_plansB | List all action plans |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 15 tools
The tools have clear distinctions in some areas, such as thinking session management (start_thinking, think, revise, conclude) and plan execution (execute_batch, execute_single, execute_plan). However, there is overlap between execute_batch and execute_single (with execute_single described as a convenience wrapper for execute_batch), and think_and_do combines thinking and execution, which could cause confusion with the separate think and execute tools. The descriptions help clarify, but some ambiguity remains in the execution and planning workflows.
Most tools follow a consistent snake_case pattern with descriptive verb_noun naming, such as start_thinking, list_sessions, and generate_plan. The main deviation is 'branch', which uses a single noun without a verb, and 'conclude', which is a verb alone. Overall, the naming is predictable and readable, with only minor inconsistencies.
With 15 tools, the count is well-scoped for the server's purpose of cognitive reasoning and execution. The tools cover a complete workflow from starting a thinking session to generating and executing plans, with each tool serving a distinct role in the process. This number is appropriate and avoids being too thin or heavy for the domain.
The tool surface provides comprehensive coverage for the cognitive reasoning domain. It includes tools for managing thinking sessions (start, add thoughts, revise, conclude), generating and managing action plans (generate, list, get), and executing operations (single, batch, plan). There are no obvious gaps, and the tools support a full lifecycle from problem analysis to execution without dead ends.