# Dexto documentation

> Official documentation for using Dexto as a durable agent-work control plane or integrating it
> through its documented REST APIs and OAuth-protected MCP server.

Start with the guide that matches the user's goal, then follow links for detail. The signed-in
`/api/*` routes listed in the API reference accept API keys; other dashboard routes are not an
external contract. Account setup, OAuth, connections, spending, and consequential actions require
explicit user authorization.

## Agent-native integration

- [Dexto integration skill](https://www.dexto.ai/skill.md): Short decision tree for agents helping
  users discover, set up, or integrate Dexto.
- [Connect via MCP](https://www.dexto.ai/docs/mcp/): Remote MCP setup for Claude, ChatGPT/Codex,
  Cursor, OpenClaw, and Hermes.
- [Codex CLI bootstrap](https://www.dexto.ai/skill.md#codex-cli-bootstrap): Agent-approved
  client setup followed by user-completed browser OAuth.
- [REST API reference](https://www.dexto.ai/docs/platform/api/): OpenAI-compatible `/v1` model
  endpoints and API-key REST service contracts.

## Dashboard workspace fallback

- [Welcome to Dexto](https://www.dexto.ai/docs/): Product overview and start paths.
- [Quickstart](https://www.dexto.ai/docs/quickstart/): Run a first clear agent task.
- [Agents](https://www.dexto.ai/docs/features/agents/): Create and configure an AI teammate.

## Platform context

- [Dexto Cloud](https://www.dexto.ai/docs/platform/): Durable runtime, services, control plane,
  and interface selection.
- [Agent primitives](https://www.dexto.ai/docs/platform/primitives/): Runtime, execution,
  connections, context, and approvals boundaries.

## REST API services

- [Models](https://www.dexto.ai/docs/platform/api/models/): Inference and model discovery.
- [CPU sandboxes](https://www.dexto.ai/docs/platform/api/sandboxes/): Sandbox lifecycle, processes, and files.
- [Search and fetch](https://www.dexto.ai/docs/platform/api/search/): Web search and page text, charged per request.
- [Files](https://www.dexto.ai/docs/platform/api/files/): Your Cloud project's file store, charged per stored GB-day.
- [Media](https://www.dexto.ai/docs/platform/api/public-media/): Media models, estimates, and generations.
- [Sites](https://www.dexto.ai/docs/platform/api/sites/): Sites, deployments, and structured Site data.
- [Runtime Artifacts](https://www.dexto.ai/docs/platform/api/artifacts/): Upload, download, and export artifacts.
- [Agents](https://www.dexto.ai/docs/platform/api/agents/): Define agents and run them for your own users.
- [Build agents by capability](https://www.dexto.ai/docs/platform/api/agents-by-capability/): Five agents, from chat to media, with what each costs.
- [Capabilities](https://www.dexto.ai/docs/platform/api/capabilities/): Discover and call capabilities.
- [Marketplace](https://www.dexto.ai/docs/platform/api/marketplace/): Browse published listings.

## Product capabilities

- [Services](https://www.dexto.ai/docs/models/): Models, media, and compute services.
- [Connections](https://www.dexto.ai/docs/features/connections/): User-approved third-party tools.
- [Approvals](https://www.dexto.ai/docs/features/approvals/): Review consequential work.
- [Claude Code](https://www.dexto.ai/docs/features/claude-code/): Run Claude Code with your own Claude account.
- [Projects](https://www.dexto.ai/docs/features/projects/): Shared task context and sources.
- [Skills](https://www.dexto.ai/docs/features/skills/): Reusable agent guidance.
- [Apps](https://www.dexto.ai/docs/features/apps/): Custom user-facing or agent-facing software.
- [Sites](https://www.dexto.ai/docs/features/sites/): Hosted agent-ready websites.
- [Loops](https://www.dexto.ai/docs/features/loops/): Recurring agent work.

## Optional

- [Dexto root agent index](https://www.dexto.ai/llms.txt): Website-wide discovery map.
- [Discover](https://www.dexto.ai/discover/): Public catalog of integrations and templates.