Getting started
Choose what to build next
Pick the next thing to give your agent, from files and tools to background work and billing.
Your agent can chat with your users. Pick what you want it to do next and start on that page. Each one works with the agent, endpoint, and chat UI you already have.
Common next steps
| I want my agent to... | Start here |
|---|---|
| Work with files and run commands | Workspaces |
| Follow reusable instructions only when they apply | Skills |
| Call my APIs or other services | MCP tools or built-in tools |
| Ask a person before a risky tool call | Tool approvals |
| Remember facts across conversations | Memory |
| Return typed JSON instead of prose | Structured output |
| Run with nobody watching, once or on a schedule | Tasks and schedules |
| Answer in Slack or Telegram | Chat integrations |
| Track token usage per user, or bill users for it | Usage and quotas and monetization |
| Run from my terminal or a CI job | CLI |
Before you go to production
Read security and credentials to handle API keys and provider keys safely, and limits and reliability to plan for product limits, retries, and cancellation.
Next
- TypeScript SDK or Python SDK for every method, field, and error.
- API reference if neither SDK fits your runtime.