Provides access to OpenAI models (e.g., GPT-*) through the gm inference gateway, allowing agents to send prompts and receive completions.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@goodmorning-mcp Ask claude-fable-5 to explain quantum computing"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
An MCP (Model Context Protocol) server that exposes models available on the gm inference gateway — a Bittensor subnet product — as callable tools, so a coding agent (Codex, Claude Code) can call gm models by name and list what's available.
npx -y goodmorning-mcpRelated MCP server: KappaML MCP Server
Node >= 22.
Sign up at saygm.com to get a GM_API_KEY.
Variable | Required | Default | Notes |
| yes | — | gm API key used as the bearer token for all requests |
| no |
| gm gateway base URL |
Add to ~/.codex/config.toml:
[mcp_servers.gm]
command = "npx"
args = ["-y", "goodmorning-mcp"]
env = { "GM_API_KEY" = "sk-...", "GM_BASE_URL" = "https://api.saygm.com/v1" }Add to .mcp.json:
{
"mcpServers": {
"gm": {
"command": "npx",
"args": ["-y", "goodmorning-mcp"],
"env": {
"GM_API_KEY": "sk-...",
"GM_BASE_URL": "https://api.saygm.com/v1"
}
}
}
}gm_askAsk a specific gm model a prompt and get back its full response text.
Param | Type | Required | Notes |
| string | yes | A gm catalog model id (see |
| string | yes | The user prompt |
| string | no | Optional system prompt |
| integer | no | Max output tokens (default 16384). Raising it increases gm's upfront credit hold; OpenAI-surface models are uncapped when it's unset. |
gm_ask auto-routes across the gm gateway's three API surfaces based on
the model's api_shapes in the gm catalog, so claude-fable-5, gpt-*,
kimi-k3, and gemini-* all work through the same tool:
Surface | Models | Endpoint |
OpenAI ( | e.g. |
|
Anthropic ( | e.g. |
|
Gemini ( | e.g. |
|
Models whose only surface is unsupported (e.g. responses-only) throw a
clear error naming the model and its api_shapes.
Non-streaming: the call blocks until the model finishes, then returns the full text in one shot. If the model returns empty content, the tool throws a clear error naming the model. The client applies a 5-minute timeout to accommodate long-reasoning models.
gm_list_modelsNo params. Returns the ids of models that are currently available and
served through a supported surface, as a newline-separated, sorted string,
for use as the model argument to gm_ask.
Once the server is configured, prompt your agent in natural language. It
will call gm_list_models to discover valid model ids, then gm_ask to
query one. For example:
Use gm to check the last response with kimi-k3.
Ask gm_list_models what's available, then get a second opinion from claude-fable-5, gpt-5.6, and gemini-3.1-pro-preview.
Model ids come and go with the gm catalog, so call gm_list_models rather
than hardcoding one.
This server cannot be installed
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Gateway between LLM agents and world data through eight tools and a bundled endpoint catalog.
Discover, inspect and run 63,000+ agent tools from one balance. Pay per call, no subscriptions.
The OpenRouter for tools. One MCP connection gives any AI agent 254 hosted tools, pay per call.
Pay-per-use tool marketplace for AI agents. Search, price-check, and call APIs via MCP.
