Orchestration FAQs

Question 1

Can Orchestration support interactive agent sessions?

Accepted Answer

Absolutely. Orchestration enables interactive agent sessions and input provision using `continue_run` when an agent run requires input or authentication, facilitating complex multi-step workflows.

Question 2

Does Orchestration store agent run data and artifacts?

Accepted Answer

Yes, it offers persistent storage for all run data, events, session information, and detailed event artifacts, including raw payloads, ensuring traceability and auditability of agent activities.

Question 3

What types of coding agents can Orchestration manage?

Accepted Answer

It can manage local instances of Codex and Claude Code, as well as remote A2A-compatible agents. This allows for flexible subagent deployment while maintaining a stable top-level API.

Question 4

How does Orchestration manage agent lifecycle?

Accepted Answer

Orchestration provides a comprehensive set of tools, including `spawn_run` to start agents, `poll_events` to monitor progress, `continue_run` for interactive sessions, and `cancel_run` to terminate them.

Question 5

What is Orchestration?

Accepted Answer

Orchestration is a TypeScript server that provides a unified interface for launching, managing, and tracking external AI coding agent runs, supporting diverse backends like Codex, Claude Code, and remote A2A agents.

Orchestration

Orchestration

主な機能

ユースケース

主な機能

ユースケース