Advanced Agent Examples
The curated advanced examples focus on workflows built on top of
ChatToolAgent. For longer-running interactive assistants, prefer the
AgentHarness examples in examples/agents/context_and_harness/.
Representative examples in the repo:
examples/agents/advanced/example_user_loop.pyexamples/agents/advanced/example_user_loop_streaming.pyexamples/agents/advanced/structured_output_agent.pyexamples/agents/advanced/output_knowledge_graph.pyexamples/agents/pipeline/example_flow_control_pipeline.pyand itsflow_control_pipeline.json— a workflow whose shape (map, parallel, loop, conditional) and endpoints both live in the JSON
What these examples cover:
- stateful user loops on top of the cleaned message and provider APIs
- structured output flows
- small orchestration patterns that sit above the base chat/tool agent
Notes:
- OCR-specific examples were intentionally retired from the maintained example surface.
- Legacy
AdvancedAgent/advanced_agent.AgentConfigexamples were removed from the maintained examples tree. Use older release tags or source archives for historical compatibility references. See Removed Legacy APIs for replacement guidance.