On 2026-11-05 DigitalOcean launched Cuttlefish, a native iPhone app for Managed Agents (Public Preview): follow live sessions and tool calls, approve or deny full commands or diffs, require Face ID/passcode for high-risk actions, and steer, pause, checkpoint, or roll back. Sessions run in cloud microVMs that keep going when the phone is locked or offline. With Box as launch partner, Box Mount (Private Preview) syncs a Box folder into the agent sandbox under Box permissions and version history.
Key Takeaways
- ✓Official: @digitalocean post 2026-11-05 14:11 UTC; blog introducing-cuttlefish; App Store, Public Preview
- ✓Live sessions/reasoning/tool calls; approve full command, path, or diff; Face ID/passcode for high-risk; steer, pause, resume, stop
- ✓In-app previews of sites/docs; browse workspace files; checkpoints with fork/rollback
- ✓Supervises harnesses including Claude Code, Codex CLI, OpenCode, OpenHands, Hermes, LangGraph, CrewAI
- ✓Box Mount (Private Preview): sync a Box folder into the sandbox; edits/deletes write back with Box permissions and version history
Key Decision Metrics at a Glance
Heavy Claude Code use: compare subscription limits and API bills
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Core Background & Industry Pain Points
Long-running coding agents often stall on high-impact steps that need a human—migrations, deletes, outbound calls—while developers are away from a keyboard. DigitalOcean Managed Agents (Public Preview) already run sessions in cloud-isolated sandboxes; on 2026-11-05 DigitalOcean shipped Cuttlefish so the human-in-the-loop can approve from an iPhone instead of returning to a stalled run.
Architecture Highlights & Internals
Cuttlefish is a native iOS client for Managed Agents. Execution and working state live in DigitalOcean-hosted microVMs, so sessions continue when the phone is locked, offline, or the app is closed. The app streams replies, reasoning, and tool calls; approvals show the full command, file path, or diff—not a summary—and high-risk requests require Face ID or a passcode. Users can steer mid-run (queued messages), pause/resume/stop, and create checkpoints to fork or roll back. Official messaging centers on choice, control, and collaboration: one app can supervise many harnesses rather than locking teams to a single vendor agent.
Authoritative Benchmarks & Measured Scores
No official benchmarks or latency numbers were published. This is a product/HITL release, not a model-score story. Engagement metrics are distribution context only, not quality scores.
Developer Hands-on Guide
With a DigitalOcean account and Managed Agents access, install Cuttlefish from the App Store, sign in via OAuth, and start a session with Claude Code, OpenCode, Codex CLI, or another supported harness. For enterprise files, connect Box under Customize > Tools > Box (Box Mount remains Private Preview). Before production use, tighten permission rules, network, and secrets, and keep high-risk tools on Ask; phone approvals complement—not replace—desktop review and change control.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.