Use BiOS from VS Code
GitHub Copilot agent mode
VS Code supports MCP natively in GitHub Copilot agent mode (VS Code 1.99 and newer). Its config format has one extra nicety: an inputs block that prompts for your API key and stores it securely instead of writing it into the file.
Before you start
You need a BiOS API key. In your dashboard, open API Keys, click Create API Key, pick a permission preset, and copy the sk_live_ key. It is shown only once.
Node.js 18 or newer must be installed; the server runs with npx, so there is nothing else to install.
Setup
- 1
Create .vscode/mcp.json
Add this file to your project. Note the top-level key is
servers(not mcpServers) in VS Code:.vscode/mcp.json{ "inputs": [ { "type": "promptString", "id": "usfbios-key", "description": "BiOS API key (starts with sk_live_)", "password": true } ], "servers": { "usfbios": { "type": "stdio", "command": "npx", "args": ["-y", "@usfbios/mcp-server"], "env": { "USFBIOS_API_KEY": "${input:usfbios-key}" } } } }VS Code asks for the key the first time the server starts and remembers it securely. The file itself is safe to commit.
- 2
Start the server
VS Code shows a Start link directly above the server entry in mcp.json, or use the Command Palette: "MCP: List Servers", then start usfbios. Enter your API key when prompted.
- 3
Use it in agent mode
Open the Chat view, switch the mode picker to Agent, and click the tools icon to confirm the BiOS tools are enabled. Then ask for what you need in plain English.
Check that it works
- The tools icon in agent mode lists usfbios tools such as
create_training_jobandsearch_models. - Ask "Recommend a GPU for fine-tuning Qwen 32B with QLoRA" and confirm a
get_recommended_gpucall.
Try these prompts
“Search for Llama models under 13B parameters and recommend a GPU for LoRA fine-tuning”
“Upload ~/data/support-qa.jsonl and start an SFT job on Llama 3.1 8B with LoRA”
“What is the status of my running training jobs? Show the loss curve for the latest one”
“Deploy my finished fine-tune and give me the endpoint URL”
Your assistant has 38 BiOS tools in total. The tool reference documents each one.
Troubleshooting
Nothing happens when the chat should call a tool
Confirm the Chat view is in Agent mode; ask mode does not run tools. Also check MCP support is enabled in Settings (chat.mcp.enabled).
I mistyped the API key at the prompt
Run "MCP: List Servers" from the Command Palette, stop usfbios, then start it again. VS Code prompts for the key on the next start.
The server never starts, or logs show "npx: command not found"
The MCP server runs on Node.js. Install Node 18 or newer from nodejs.org, restart your client, and try again. Run npx -y @usfbios/mcp-server in a terminal to confirm it starts (it waits for input; press Ctrl+C to exit).
Tools appear but every call fails with an authentication error
Check that USFBIOS_API_KEY is set to a key that starts with sk_live_ and has not expired. Keys are shown only once at creation, so if in doubt create a fresh one under Dashboard, then API Keys.
A tool call fails with a permission error
Your key may be scoped too narrowly. Ask the assistant to call introspect_api_key to see exactly which scopes the key has, then create a key with the preset you need (Read Only, Training, or Full Access).
Frequently asked questions
Why is the format different from Cursor and Claude?
VS Code uses servers plus an optional inputs block, while most other clients use mcpServers. The command, args, and env inside are identical.
Can I set it up once for all projects?
Yes. Run "MCP: Add Server" from the Command Palette and choose the user (global) location instead of the workspace, or add the same block to your user-level mcp.json.
Use BiOS from other clients
The same server works everywhere MCP does. More setup guides:
Terminal coding agent by Anthropic
Claude app for macOS and Windows
claude.ai in the browser and the mobile apps
ChatGPT and the OpenAI Codex CLI
AI code editor
AI code editor with Cascade agent
Open-source VS Code agent extension
High-performance editor with Agent Panel
Autonomous software engineer by Cognition
Google terminal coding agent