microsoft/autogen
A programming framework for agentic AI
How autogen is put together
AutoGen is a framework for building multi-agent AI applications, structured as a layered, extensible system with a Python monorepo (uv workspace) and a parallel .NET implementation. The core provides event-driven message passing and local/distributed agent runtimes (including a gRPC cross-language runtime), with higher layers offering an opinionated AgentChat API and an Extensions API for LLM clients (OpenAI, Azure, Anthropic, Ollama, Gemini) and capabilities like code execution and web/file surfing. Developer tools include AutoGen Studio (a no-code GUI with a Gatsby/React frontend and Python backend) and AutoGen Bench for benchmarking. Data flows as typed messages routed to agent handlers via a runtime, with components serialized to/from portable ComponentModel configs. The project is now in maintenance mode.
Languages
Frameworks
Datastores
Infrastructure
Major components
autogen-core
Foundational interfaces, message passing, event-driven agent runtime (local and distributed), and component config serialization.
autogen-agentchat
Opinionated higher-level API for agents and teams supporting common multi-agent patterns like two-agent and group chats.
autogen-ext
First/third-party extensions: LLM clients (OpenAI, Azure, Anthropic, Ollama, Gemini), code executors, web/file surfers, MCP, and the gRPC runtime.
autogen-studio
No-code GUI (Gatsby/React frontend + Python backend) for prototyping and running multi-agent workflows.
agbench
CLI benchmarking suite for evaluating agent performance, using Docker and HuggingFace/OpenAI.
magentic-one-cli
State-of-the-art multi-agent team for web browsing, code execution, and file handling built on AgentChat/Extensions.
dotnet (AutoGen .NET)
C# implementation of AutoGen with agents, group chats, middleware, and model connectors (OpenAI, Anthropic, Gemini, Mistral, Ollama, SemanticKernel).
protos / gRPC runtime
Protobuf definitions (agent_worker, cloudevent) enabling the cross-language distributed agent worker runtime.
No digest this week
1 weeksWeek of
Dependencies and code review
Dependency advisories
Security Watch
No known advisories across 0 scanned dependencies.
No known advisories in the scanned dependencies.
Code review
GET/PUT/DELETE /api/sessions/ and /api/sessions/{session_id} (autogenstudio/web/routes/sessions.py: list_sessions, get_session, update_session, delete_session, list_session_runs)
Cross-tenant read, modification, and deletion of any user's chat sessions, runs, and messages (horizontal privilege escalation / IDOR).
Fix: Stop accepting `user_id` from the query/body. Derive it from the authenticated principal, e.g. add a dependency `user: User = Depends(get_current_user)` and use `user.id` in every db filter, rejecting or ignoring any client-supplied user_id.
Get this report every week for your repos.
GitZoid learns each repo, reports what changed, and flags what needs attention. One flat price for the whole team.
$19 a month, flat · First 10 outputs free · No card required