Agents

Agent examples cover the loop end-to-end through scenarios you’d build yourself. Sources under examples/agents/.

Name	Scenario	Source
`agents_react_agent`	The canonical ReAct agent — model decides which tool to call to answer a question.	src
`agents_conversational`	Multi-turn recall test — user gives their name in turn 1, asks for it in turn 2.	src
`agents_multi_agent`	Sequential handoff — one agent’s reply becomes the next agent’s input.	src
`agents_planning`	Planning middleware — the agent drafts a plan before tool selection.	src
`agents_streaming_events`	Stream lifecycle events from a running agent — what wires into a UI’s progress display.	src

How to run

COGNIS_PROVIDER=ollama COGNIS_OLLAMA_MODEL=llama3.1 \
  cargo run -p cognis-examples --example agents_react_agent

The mental model.

Sequential / Supervisor / ParallelVote / RoundRobin strategies.

Plug memory into any of these agents.

A worked multi-agent application.