Early Access. Commands and behavior can change between releases.
Sub-agents and teams
Run agents in parallel, then watch and steer them live.
Workflows
A short script that runs agents over a list and combines their results.
Monitor
Watch a long command in the background and get its output as notifications.
Trace viewer
Browse your session logs with /trace. Read-only.
Project context
Your project’s instructions, loaded from AGENTS.md or CLAUDE.md into your agents’ prompts.
MCP servers
Connect your own MCP servers. The model calls their tools directly.
System agents
The Orchestrator, Architect, Developer and LLM-Whisperer, and the commands that call them.
Orchestrator and agents
The software-engineering roles and how they work together.
Onboard your project
Set up project context, engineering rules and an optional board.