GitHub Copilot

Copilot CLI through itr's hook, Copilot Chat with a model on the Intertrace edge, and MCP for the coding agent.

Copilot CLI

itr connect copilot installs a pre-tool hook: every tool call Copilot CLI makes is decided by the enforcer first, and a call that needs a person waits for one.

itr connect copilot
itr doctor

Copilot Chat in VS Code

Mint itr_rt_* in Settings → Runtime API keys. Keep upstream secrets in the Provider vault; clients should not need vendor sk- keys.

Add a Custom Endpoint model (Chat: Manage Language Models → Add Models → Custom Endpoint) that points at the edge with your runtime key. Chat and agent mode with that model then go through Intertrace.

[
  {
    "name": "Intertrace",
    "vendor": "customendpoint",
    "apiKey": "${input:intertraceRuntimeKey}",
    "apiType": "chat-completions",
    "models": [
      {
        "id": "gpt-4o-mini",
        "name": "GPT-4o mini via Intertrace",
        "url": "https://intertrace.fly.dev/v1/chat/completions",
        "toolCalling": true
      }
    ]
  }
]

Copilot coding agent and MCP

In the repository's Settings → Copilot → MCP servers, add Intertrace's /mcp with an Agents secret named COPILOT_MCP_INTERTRACE_KEY. Every tool call is decided at the edge. See MCP.

{
  "mcpServers": {
    "intertrace": {
      "type": "http",
      "url": "https://intertrace.fly.dev/mcp?server=YOUR_SERVER_KEY",
      "headers": { "Authorization": "Bearer $COPILOT_MCP_INTERTRACE_KEY" },
      "tools": ["*"]
    }
  }
}