OpenRouter Fusion MCP Server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| OPENROUTER_API_KEY | Yes | Your OpenRouter API key (inference key) with credit. | |
| OPENROUTER_FUSION_PRESETS | No | Legacy: a JSON map of preset names to config objects. | |
| OPENROUTER_FUSION_PERSO_CONFIG | No | Legacy: a single custom config object for 'perso' preset. | |
| OPENROUTER_FUSION_DEFAULT_REASONING | No | Default reasoning effort for fusion configs (e.g., high, medium). |
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": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| fusion_listA | List the available Fusion configurations — name, label, panel (analysis_models), judge (orchestrator/'Fuse with'), reasoning_effort, temperature, description. Call this FIRST when the user asks to 'use Fusion' without naming a config: show the list, let them choose (quality, budget, or a custom one), then run fusion_start with preset:''. |
| fusion_startA | Start an OpenRouter Fusion deliberation in the BACKGROUND and return a job_id immediately (never times out), then call fusion_result with the job_id. Pick a config with preset:'' — call fusion_list to see them (quality, budget, + custom configs). Default: quality. reasoning_effort and temperature default to the config's; override per call if needed. |
| fusion_resultA | Fetch the result of a fusion_start job. Long-polls up to ~45s: returns the synthesized answer when ready, or {status:'running'} (call again with the same job_id). Fast per call, so it never hits the client timeout. |
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 3 tools
Each tool has a clearly distinct purpose: listing configurations, starting a job, and fetching results. No overlap in functionality.
All tools follow a consistent 'fusion_' prefix with a verb-like suffix (list, start, result), making the pattern predictable and clear.
Three tools cover the essential workflow of using Fusion: discover configs, start a job, and retrieve results. No unnecessary tools.
Core lifecycle is covered, but missing a tool to cancel or abort a running job, which could be a minor gap in error recovery.