Legion MCP server

Legion

"I am Legion, for we are many."

An MCP-native model council. Legion exposes LLMs as individual tools and orchestrates them into debates, juries, blind panels, private refinement gauntlets, workshops, and custom multi-model deliberations.

Every model is reached through the OpenAI Responses API wire format. Use OpenAI or Azure directly, route other providers through a compatible gateway (such as a LiteLLM proxy), and configure the entire council through hot-reloadable files.

Contents

• How it works
• Design decisions
• Requirements
• Setup
• Configuration
• Logging
• Run
• Try it
• Use in VS Code
• Deploy

How it works

flowchart LR
   AI[Calling AI] -->|claude / gpt / gemini …| Legion
   Legion -->|Responses API| GPT[OpenAI / Azure — direct]
   Legion -->|Responses API| GW[Gateway e.g. LiteLLM]
   GW --> Claude & Gemini & Llama

claude). Each accepts a prompt plus optional context, role, system, temperature, and maxTokens.

• One tool per model, named after the slugified model name (e.g. Claude →

multi-round discussion, visibility modes, and synthesis — and returns each answer separately. See Presets for the orchestration options.

• A quorum tool fans one prompt out to several models — with roles,

each exposed as its own tool.

• Presets are named, pre-staffed councils (debate, courtroom, code review, …),

Logging goes to stderr (safe for stdio).

• Identity and telemetry ride in structuredContent, not the answer text.

Design decisions

model list. Legion speaks one wire format; models that don't speak it natively go through a gateway. Supporting a new model requires no change here.

• No provider adapters. There is no provider-specific code and no built-in

directory is re-read per request, so no rebuild or restart.

• Models are config, not code. Adding a model means adding a JSON file. The

with its own description, rather than a single tool with a model parameter. The quorum tool covers the ad-hoc multi-model case, and each preset in config/presets/ is exposed as its own enforced, pre-staffed council tool.

• One tool per model. Each model appears to the calling AI as its own tool

persisted, so there is no database and no conversation state to manage.

• Stateless. Every call is one-shot with store: false. Nothing is

dependencies.

• Small. A few hundred lines of TypeScript, one bundled output file, six

Requirements

a gateway such as LiteLLM for models that need bridging)

• Node.js 24+
• At least one OpenAI-Responses-compatible endpoint (a provider API directly, or

Setup

From npm — no clone, no build:

npx legion-mcp

From a clone:

npm install
copy .env.example .env   # then edit .env

Configuration

All configuration lives in a config/ directory. The bundled defaults are always the base layer; a config/ folder in the current working directory is overlaid on top of them, per file:

file overrides the bundled file of the same name; a local-only file is added; every bundled file you don't touch stays. So dropping in one config/presets/refine.json overrides just that preset — the other bundled presets remain.

• Directory resources (models/, roles/, presets/, tools/): a local

per key — bundled < local. A partial local file overrides only the keys it sets.

• Single-file text (prompts.json, errors.json, schema.json): merged

• description.md: local wins whole if present, else bundled.

The overlay can override or add, but not delete a bundled entry. To turn off bundled presets you don't want, use DISABLE_PRESETS (see below).

Shortened. The full README is on GitHub.