# Frameworks & middleware

Agent frameworks and federated-learning frameworks.

- [OpenAI Agents SDK & Responses API](openai-agents-sdk.md) — Multi-tool/multi-agent orchestration, built-in tools, streaming, observability.
- [Semantic Kernel](semantic-kernel.md) — Middleware translating model requests into existing functions/plugins.
- [AutoGen](autogen.md) — Conversational single- and multi-agent programming framework.
- [Flower](flower.md) — Framework-agnostic federated AI: simulation-to-deployment, secure agg, DP, audit, OIDC.
- [NVIDIA FLARE](nvidia-flare.md) — Production-oriented FL SDK: admin, monitoring, access control, confidential computing.
- [Substra](substra.md) — Production FL for regulated health/biotech; Kubernetes, auditable logs.
- [FATE](fate.md) — Industrial-grade FL with HE/MPC secure computation.
- [PySyft](pysyft.md) — Privacy-preserving remote analysis on data that stays put.
- [FedML / TensorOpera](fedml.md) — Scalable distributed training/deployment across GPUs, cloud, edge, smartphones.
- [TensorFlow Federated](tensorflow-federated.md) — Research framework for custom federated computations.
- [OpenFL](openfl.md) — Open Federated Learning library (LF AI & Data).
- [Flotilla](flotilla.md) — Research system for large-scale, asynchronous, fault-tolerant federations.
- [Context Hub](context-hub.md) — Andrew Ng open content/skills registry for coding agents.