Generated from commit be57775 ·
How Agent works, from the code.
Each page is built from the source on every release: the screens, the operations behind them and the guide that explains them.
60 features · 614 API operations · 33 MCP tools
Getting started
- Accept an invitationSign in with Google, GitHub, a code sent to your email, or a password. Sign-up asks for your name, your organization and your role.
- Device sign-inThe runner is a small program on your own computer. It starts your own Claude Code or Codex, under your own sign-in, for work that you asked for.
- First owner setupCreates the first owner and organization on a new installation.
- OnboardingThe setup wizard has seven pages. It prepares your CEO to start work. You can stop at any time and come back later.
- Sign inSign in with Google, GitHub, a code sent to your email, or a password. Sign-up asks for your name, your organization and your role.
- Sign upSign in with Google, GitHub, a code sent to your email, or a password. Sign-up asks for your name, your organization and your role.
Command
Work
- PR trialsCompare past pull requests by result, checks, time and estimated cost.
- BoardSee work by status, owner and priority.
- FlowSay what you need. Triage employees sort, rank and assign it.
- DeliverySee what shipped each day, lead time and who owns each stage gate.
- Day replayPlay back every ticket that moved on one day.
- SpecsWrite plans together and trace each requirement to its evidence.
- ApprovalsApprove or deny actions that need your permission.
- EscalationsResolve questions with the options and evidence workers found.
- InboxThreads, requests and mentions between people and workers.
- AutomationsWork your team does on its own: on a schedule, once later, or when something happens.
- OutreachControl who may contact people outside your organization.
- IntakeBring work in through trackers, schedules and events.
- New workA template fills in the objective, what done means and who owns it. Agent has ten built-in templates for common office work.
- Work itemA work item is one piece of work with an owner, a state, a plan, checks and evidence. Each one has a key such as AGT-12.
Organization
- OrganizationSee teams, reporting lines, workload and spend.
- Company plannerPlan your company, preview workers and run repeatable workflows.
- WorkersSee your workers and their current work.
- ResourcesManage resources, owners and access requests.
- OversightReview risks, failures and paused work with their evidence.
- RolesDefine worker roles and their responsibilities.
- Levels & EvaluationsReview worker results and proposed level changes.
Knowledge
- Repository AtlasExplore each repository through its files, rules and architecture.
- Company dataConnect your tools, see how your records connect and ask questions with sources.
- MemoryReview what workers remember and where it came from.
- Operating ModelReview learned conventions before you accept them.
- DocumentsKeep sources that workers can search and cite.
- LessonsReview lessons from finished work.
- Examples & ProceduresReuse approved examples and review proposed procedures.
- Your personaReview your private work preferences. Workers use them after you accept them.
Platform
- IntegrationsConnected tools and which workers may use them.
- Brain & ModelsChoose models and check provider health.
- Model accountsManage model accounts, credentials, budgets and spend.
- DecisionsReview how workers make decisions and how well they perform.
- PoliciesSet what workers may do and when they need approval.
- Rules & PreferencesSet work rules and preferences. Review their sources here.
- Usage & spendSee worker time, tokens, tools and model costs. Set spend limits.
- Audit LogReview the permanent record of actions and changes.
- MembersPeople in this organization and their roles.
- API tokensManage access tokens for your tools and agents.
- SettingsOrganization name, defaults and preferences.
- Billing settingsThe Billing page shows plans and invoices only when the operator of your Agent server has set up billing. Until then, prices and checkout stay off.
- ConnectionsAdd Agent to Claude, ChatGPT, Claude Code or Codex. The model runs there on your plan and uses Agent's tools to read and assign work.
- Enterprise identityEnterprise identity connects your identity provider with OIDC or SAML, provisions people with SCIM, and can require single sign-on for your email domain.
- RunnersOn macOS, work on your runner runs in a sandbox. It can write only its own task folder and reach only the model vendor and allowed package registries.
- TriggersAgent can start work on your own Claude routine or ChatGPT Workspace Agent while you are away. You save its trigger in Agent.
Admin
Help
Code and integrations
For agents and tools
The same docs are JSON: index.json lists every feature, /docs-data/features/<slug>.json holds one feature, and api.json and mcp.json hold the API and the MCP tools. Members see each feature's files, symbols and tests in its code view.