Agent Lifecycle

When you run an agent, it transitions through several conceptual phases to ensure safe and accurate execution.

1. Planning: The agent decides which tools to call.
2. Executing: The agent executes the tools.
3. Awaiting Approval: If a tool is gated, execution pauses until human confirmation.
4. Verifying: Post-execution checks and guardrails run.
5. Done: The turn is complete.

More content coming soon.