n8n-nodes-github-copilot-agent-tool

GitHub Copilot nodes for n8n with AI Agent Tool support. Access GPT-4, Claude, Gemini and more using your existing Copilot subscription.

Nodes

| Node | Description | |---|---| | GitHub Copilot Agent Tool ⭐ | AI-agent-ready tool with dynamic model selection, structured output, and automatic fallback | | GitHub Copilot Chat API | Direct chat completions + model listing | | GitHub Copilot Chat Model | Multi-turn chat with full tool/function-calling | | GitHub Copilot Embeddings | Text embeddings for semantic search and RAG | | GitHub Copilot OpenAI | Full OpenAI-compatible surface with tool support | | GitHub Copilot CLI | Shell command suggestions via gh copilot CLI | | GitHub Copilot Auth Helper | Browser-based OAuth Device Flow authentication | | GitHub Copilot Test | Credential verification and diagnostics |

Installation

Via n8n UI: Settings → Community Nodes → Install → n8n-nodes-github-copilot-agent-tool

Or with npm:

npm install n8n-nodes-github-copilot-agent-tool

Authentication

Option A — GitHub Copilot CLI (local n8n)

gh auth login
gh auth token   # copy this token

Paste the token into the GitHub Copilot API credential in n8n.

Option B — OAuth Device Flow (hosted n8n)

1. Add a GitHub Copilot Auth Helper node → Start Device Flow
2. Visit the URL shown and enter the user code
3. Add a second Auth Helper node → Poll for Token
4. Use the returned access_token in the GitHub Copilot API credential

GitHub Copilot Agent Tool

Drop this node into any AI Agent workflow as a tool.

Key parameters

| Parameter | Default | Description | |---|---|---| | Model | claude-sonnet-4-5 | AI model — supports expression bindings | | Prompt / Query | — | User message | | System Prompt | helpful assistant | System instructions | | Temperature | 1 | 0 = deterministic, 2 = creative | | Max Tokens | 2048 | Response length cap | | Agent Mode | false | Adds concise/structured instructions to system prompt | | Fallback Model | — | Model to use if primary returns 404 | | On Error | return | return structured error object or throw |

Dynamic model selection

={{ $json.preferred_model || "claude-sonnet-4-5" }}

Structured output

{
  "success": true,
  "model_used": "claude-sonnet-4-5",
  "response": "Here is the answer...",
  "tokens_used": 342,
  "latency_ms": 1204,
  "metadata": {
    "stop_reason": "stop",
    "finish_details": {},
    "request_id": "chatcmpl-abc123"
  }
}

Error codes

| Code | Trigger | |---|---| | INVALID_CREDENTIALS | 401 / 403 — bad or expired token | | UNSUPPORTED_MODEL | 404 — model not found | | RATE_LIMITED | 429 — quota exceeded | | CONTEXT_LENGTH_EXCEEDED | 400 — prompt too long | | NETWORK_ERROR | ECONNREFUSED / ETIMEDOUT | | UNKNOWN | Other errors |

Supported Models

gpt-4 · gpt-4-turbo · gpt-4o · gpt-3.5-turbo · claude-sonnet-4-5 · claude-sonnet-4 · claude-3-opus · claude-3-sonnet · gemini-1.5-pro · gemini-1.5-flash · o1-preview · o1-mini · Custom

License

MIT — see LICENSE