One workspace.
Many working agents.

Hire the right agents, run them as Agent Teams, and share the files, skills, browsers, auth, and decisions every run needs.

Agent hirer Define role, runtime, skills, tools, and scope before the agent starts.
Agent teamwork Run PM, builder, reviewer, and specialist agents together in one visible flow.
Shared context Keep files, browsers, auth, decisions, and outputs attached to the workspace.
Download for macOS Start now
OneShip Agent Teams workspace showing specialized agents coordinating work and shared output.
Scheduled tasks Workspace skills Agent browsers Codex / Claude Code Workspace auth

Core benefits

Three things users should understand first.

OneShip is not another single-agent chat. It is an operating workspace for hiring agents, coordinating agent teams, and keeping shared context alive across the work.

Create Agent screen showing runtime, instructions, skills, tools, and visibility.
Role, runtime, skills, tools, and visibility live in one reviewable profile.

Agent hirer

Create the right agent before work starts.

Agent hiring should feel like defining a teammate, not wiring a framework. OneShip turns a role into a profile with runtime, tools, skills, and access you can inspect before it acts.

  • Use your own Codex or Claude Code when you want.
  • Use OneShip's managed Claude agent when cloud execution is better.
  • Let Haru help draft an agent, but keep the hiring decision visible.
Agent Teams screen showing several agents coordinating and producing shared output.
Agent Teams make planning, building, review, and shared output visible in one place.

Agent teamwork

Specialists work together, not in separate chats.

A single request can become a coordinated run: one agent plans, another builds, another researches, and another reviews before a human accepts the result.

  • Agent Teams split the work by role.
  • To-dos keep status visible across queued, running, review, and done.
  • Shared output keeps the result attached to the run.
Auth screen showing scoped provider and API-key access for agents.
Workspace auth scopes provider access once, then agents use only what they are allowed to use.

Workspace context sharing

Context belongs to the workspace, not the prompt.

Files, skills, browsers, credentials, projects, and previous decisions should not be pasted into every new conversation. OneShip keeps them shared and scoped for the people and agents doing the work.

  • Manage skills across the workspace, then attach the right skill to the right agent.
  • Give every agent its own browser for evidence and web-app work.
  • Manage auth once, without pasting keys to every collaborator.

Use cases

Where agent work becomes easier to trust.

Software
Turn a product request into planning → implementation → review → human approval, with context and files kept in the workspace.
Research
Use agent browsers to gather evidence → summarize findings → attach sources → hand a reviewable brief back to the team.
Content ops
Brief the work → draft assets → review claims → assemble the package → keep every artifact attached to the run.
Recurring ops
Schedule checks, reports, cleanup, and summaries so agents run automatically after setup instead of waiting for someone to restart the work.

Capabilities

The functional layer underneath.

Workspace operating layer

Put execution, context, access, and output under one roof.

These capabilities are the infrastructure that lets Agent Teams keep moving without rebuilding context every run.

Run

Scheduled tasks

Runs automatically after setup.

One-time, recurring, or triggered work can run without someone manually restarting the task.

Runtime choice

Use the agent stack you want.

Bring your own Codex or Claude Code, or use OneShip's managed Claude agent when cloud execution is better.

Equip

Workspace skills

Right skill, right agent.

Manage skills across the workspace and attach reusable methods to the agents that need them.

Agent browsers

Every agent gets its own browser.

Agents can research sites, use web apps, collect screenshots, and bring evidence back to the run.

Govern

Workspace auth

Stop pasting keys everywhere.

Manage provider access and API keys once, scoped to agents and collaborators in the workspace.

Shared artifacts

Work returns to the workspace.

Files, notes, tasks, and output stay attached to the run so humans can review and continue.

Desktop app

Connect your Mac for local agent work.

Install the macOS app to keep local runtimes online, bind folders, and let agents use the tools already on your machine.

Runtime Keep local agent runtimes connected and remembered.
Folders Bind source folders and working directories from your Mac.
Tools Use installed runtimes like Codex CLI and Claude Code.
Download for macOS