Developers increasingly want to build agents that can reason about code, modify files, execute commands, interact with developer tools, and work across entire repositories. While GitHub Copilot already provides a powerful coding harness for these scenarios, developers often need additional capabilities such as observability, middleware, approval workflows, enterprise governance, and integration with broader agent ecosystems.<\/p>\n
The GitHub Copilot integration in Microsoft Agent Framework brings these worlds together. You can now use GitHub Copilot’s agentic harness as the execution engine for your agents while continuing to leverage Agent Framework’s extensibility, tooling model, observability, streaming, and human-in-the-loop approval experiences.<\/p>\n
Today, we’re excited to announce that the GitHub Copilot Agent is now released and stable for both\u00a0.NET<\/strong>\u00a0and\u00a0Python<\/strong>\u00a0making it easier than ever to build production-ready coding agents using familiar Agent Framework abstractions.<\/p>\n The GitHub Copilot agent is an Agent Framework agent backed by the GitHub Copilot CLI and SDK. Copilot owns the agent loop (model calls, tool invocation, planning, and session state) while Agent Framework gives you a consistent surface for instructions, tools, streaming, middleware, observability, and human-in-the-loop approval.<\/p>\n The result: an agent with Copilot’s built-in coding-agent capabilities \u2014 shell execution, file read\/write, URL fetching, and MCP tools \u2014 wired into the same run interface as every other Agent Framework provider.<\/p>\n .NET<\/strong><\/p>\n Python<\/strong><\/p>\n Both give you streaming too: Copilot’s harness comes with the abilities a coding agent needs, and you opt in to each one through the permission handler:<\/p>\n Because every capability is gated by a permission request, the agent can only do what you explicitly allow. The handler receives each request and returns an approve\/deny decision:<\/p>\n .NET<\/strong><\/p>\n Python<\/strong><\/p>\n Configure Model Context Protocol servers \u2014 local ( .NET<\/strong><\/p>\n Python<\/strong><\/p>\n Register functions as tools alongside Copilot’s built-ins. Tools that require approval are gated through Copilot’s native pre-tool-use hook and routed to your approval handler. Wrap an .NET<\/strong><\/p>\n Python<\/strong><\/p>\nWhat is the GitHub Copilot Agent?<\/h2>\n
using<\/span> GitHub.Copilot;\r\nusing<\/span> GitHub.Copilot.Rpc;\r\nusing<\/span> Microsoft.Agents.AI;\r\n\r\n\/\/ Start a Copilot client and turn it into an AIAgent.<\/span>\r\nawait<\/span> using<\/span> CopilotClient copilotClient = new<\/span>();\r\nawait<\/span> copilotClient.StartAsync();\r\n\r\nSessionConfig sessionConfig = new<\/span>()\r\n{\r\n OnPermissionRequest = (request, invocation) =>\r\n Task.FromResult(PermissionDecision.ApproveOnce()),\r\n};\r\n\r\nAIAgent agent = copilotClient.AsAIAgent(sessionConfig, ownsClient: true<\/span>);\r\n\r\nAgentResponse response = await<\/span> agent.RunAsync(\"Summarize what this project does.\"<\/span>);\r\nConsole.WriteLine(response);\r\n<\/code><\/pre>\nimport<\/span> asyncio\r\nfrom<\/span> agent_framework.github import<\/span> GitHubCopilotAgent, GitHubCopilotOptions\r\nfrom<\/span> copilot.session import<\/span> PermissionHandler\r\n\r\n\r\nasync<\/span> def<\/span> main<\/span>() -> None<\/span>:\r\n async<\/span> with<\/span> GitHubCopilotAgent(\r\n instructions=\"You are a helpful assistant.\"<\/span>,\r\n default_options=GitHubCopilotOptions(\r\n on_permission_request=PermissionHandler.approve_all,\r\n ),\r\n ) as<\/span> agent:\r\n result = await<\/span> agent.run(\"Summarize what this project does.\"<\/span>)\r\n print<\/span>(result)\r\n\r\n\r\nif<\/span> __name__ == \"__main__\"<\/span>:\r\n asyncio.run(main())\r\n<\/code><\/pre>\nRunStreamingAsync(...)<\/code>\u00a0in .NET,\u00a0run(..., stream=True)<\/code>\u00a0in Python.<\/p>\nWhat you can do with it<\/h2>\n
Give an agent real system capabilities<\/h3>\n
\n
static<\/span> Task<PermissionDecision> PromptPermission<\/span>(PermissionRequest request, PermissionInvocation invocation<\/span>)<\/span>\r\n{\r\n Console.WriteLine($\"[Permission Request: {request.Kind}<\/span>]\"<\/span>);\r\n Console.Write(\"Approve? (y\/n): \"<\/span>);\r\n string<\/span>? input = Console.ReadLine()?.Trim().ToUpperInvariant();\r\n return<\/span> Task.FromResult(input is<\/span> \"Y\"<\/span> or<\/span> \"YES\"<\/span>\r\n ? PermissionDecision.ApproveOnce()\r\n : PermissionDecision.Reject());\r\n}\r\n<\/code><\/pre>\ndef<\/span> approve_and_log<\/span>(request, context<\/span>):\r\n if<\/span> request.kind == \"shell\"<\/span>:\r\n print<\/span>(f\"[Permission: {request.kind}<\/span>] {getattr<\/span>(request, 'full_command_text', '')}<\/span>\"<\/span>)\r\n return<\/span> PermissionHandler.approve_all(request, context)\r\n return<\/span> PermissionDecisionUserNotAvailable()\r\n<\/code><\/pre>\nExtend it with MCP servers<\/h3>\n
stdio<\/code>) or remote (http<\/code>) \u2014 to give the agent tools and data beyond the built-ins, from a filesystem server to remote services like the Microsoft Learn documentation API.<\/p>\nSessionConfig sessionConfig = new<\/span>()\r\n{\r\n OnPermissionRequest = PromptPermission,\r\n McpServers = new<\/span> Dictionary<string<\/span>, McpServerConfig>\r\n {\r\n [\"filesystem\"<\/span>] = new<\/span> McpStdioServerConfig\r\n {\r\n Command = \"npx\"<\/span>,\r\n Args = [\"-y\"<\/span>, \"@modelcontextprotocol\/server-filesystem\"<\/span>, \".\"<\/span>],\r\n Tools = [\"*\"<\/span>],\r\n },\r\n [\"microsoft-learn\"<\/span>] = new<\/span> McpHttpServerConfig\r\n {\r\n Url = \"https:\/\/learn.microsoft.com\/api\/mcp\"<\/span>,\r\n Tools = [\"*\"<\/span>],\r\n },\r\n },\r\n};\r\n\r\nAIAgent agent = copilotClient.AsAIAgent(sessionConfig, ownsClient: true<\/span>);\r\n<\/code><\/pre>\nmcp_servers = {\r\n \"filesystem\"<\/span>: {\r\n \"type\"<\/span>: \"stdio\"<\/span>,\r\n \"command\"<\/span>: \"npx\"<\/span>,\r\n \"args\"<\/span>: [\"-y\"<\/span>, \"@modelcontextprotocol\/server-filesystem\"<\/span>, \".\"<\/span>],\r\n \"tools\"<\/span>: [\"*\"<\/span>],\r\n },\r\n \"microsoft-learn\"<\/span>: {\r\n \"type\"<\/span>: \"http\"<\/span>,\r\n \"url\"<\/span>: \"https:\/\/learn.microsoft.com\/api\/mcp\"<\/span>,\r\n \"tools\"<\/span>: [\"*\"<\/span>],\r\n },\r\n}\r\n\r\nagent = GitHubCopilotAgent(\r\n instructions=\"You are a helpful assistant with filesystem and Microsoft Learn access.\"<\/span>,\r\n default_options=GitHubCopilotOptions(\r\n on_permission_request=PermissionHandler.approve_all,\r\n mcp_servers=mcp_servers,\r\n ),\r\n)\r\n<\/code><\/pre>\nAdd your own tools<\/h3>\n
AIFunction<\/code>\u00a0in\u00a0ApprovalRequiredAIFunction<\/code>\u00a0in .NET, or declare\u00a0approval_mode=\"always_require\"<\/code>\u00a0in Python.<\/p>\n\/\/ Wrap a tool in ApprovalRequiredAIFunction to gate it behind OnPermissionRequest.<\/span>\r\nAIFunction getWeather = AIFunctionFactory.Create(GetWeather);\r\n\r\nAIAgent agent = copilotClient.AsAIAgent(new<\/span> SessionConfig\r\n{\r\n OnPermissionRequest = PromptPermission,\r\n Tools = [new<\/span> ApprovalRequiredAIFunction(getWeather)],\r\n SystemMessage = new<\/span> SystemMessageConfig\r\n {\r\n Mode = SystemMessageMode.Append,\r\n Content = \"You are a helpful weather assistant.\"<\/span>,\r\n },\r\n}, ownsClient: true<\/span>);\r\n<\/code><\/pre>\nfrom<\/span> typing import<\/span> Annotated\r\nfrom<\/span> agent_framework import<\/span> tool\r\n\r\n\r\n