Real, editable workflows. Loading one creates a new project you can change freely.
Build the agent loop yourself: an LLM Call that can use tools, an Execute Tools step, and a loop back — exactly like the LangGraph quickstart.
Pull structured fields out of a message into variables, branch on them, and write a reply that uses them.
A single agent with a model and conversation memory. The best place to start.
An LLM router sends billing and technical questions to specialised agents with their own tools.
A ReAct agent that searches the web, reasons over results and writes a sourced answer.
Answers questions from your own documents: load, split, embed, store and retrieve.
Three specialists research in parallel; a synthesiser merges their findings into one report.
A supervisor delegates to a researcher and a writer, then a human approves the draft.
An assistant that uses tools from a remote MCP server — here, the public LangChain docs server.
A Deep Agent plans with todos, delegates to a researcher and critic, and follows a report-writing skill.