View as Markdown

Features

Beyond interactive chat, Crew runs work in the background, on a schedule, and across machines. This section covers everything that extends beyond a single conversation.

Subagents
Spawn parallel background agents to research, prototype, or investigate — results flow back automatically.
Scheduling
Recurring jobs, reactive monitoring, external triggers, and in-session checks.
Artifacts
Persistent output with version history, live previews, and one-click deploy to your own AWS account.
Agent backends
Choose the harness that runs new Crew sessions, including Kiro CLI, Claude Code, Kiro Agent, Codex, OpenCode, Pi, and goose.
Multi-instance
Manage several remote Crew instances from one hub. Start a chat on a connected crew (Preview).
Task Runner
Give Crew a spec and walk away — it plans, executes, validates, retries, and checkpoints.
Memory
Automatic memory that builds from conversations — preferences, history, lessons, and semantic facts.
Knowledge
A curated library of documents, folders, and URLs you feed to the agent for reference.
Snapshot & restore
Back up everything Crew knows into a portable tarball. Restore on a new machine or recover from an issue.
Browser
Let the agent operate a real browser — navigate, click, type, and fill forms — with the live view in the dashboard.
Computer Use
Drive native desktop apps through the accessibility layer. Press, type, scroll, and drag. macOS and Windows, off by default.
Workflows
Orchestrate many agents through fan-out, pipelines, and judge-and-verify stages with an authored Python script.
Page updated: November 11, 2026