# Integrations

External services Everruns connects to — sandboxes, browsers, search, messaging, and evaluation — exposed as first-class surfaces rather than bolt-ons.

[← Back to home](/)

## Why integrations are core

Execution sandboxes, browser automation, search, and evaluation tooling are what make a durable runtime useful inside a real stack. Everruns treats them as part of the core story, attached the same way as any other [capability](/platform/capabilities), so the durable execution model never changes when you add one.

## Sandboxes and execution

Isolated environments for running untrusted agent code. These connect over each provider's own API and surface as sandbox capabilities:

- **Daytona** — cloud sandbox environments via the Daytona REST API
- **E2B** — cloud sandboxes via the E2B management and runtime APIs
- **Deno** — cloud sandboxes via the Deno WebSocket sandbox API
- **Docker** — self-hosted container-based execution via the Docker Engine
- **Sprites** — persistent Firecracker microVMs with checkpoints and HTTP services
- **Cursor** — launch and manage asynchronous Cursor Cloud coding agents on GitHub repositories

See [Sandboxes](/platform/sandboxes) for how these compare and when to reach for each.

## Browser and web

- **Browserless** — cloud browser automation for screenshots, DOM access, scraping, and multi-step flows
- **Brave Search** and **DuckDuckGo** — web search and instant answers
- **Parallel** — web search, extract, and task APIs

## Messaging and discovery

- **Slack** — deploy an agent as a Slack bot through the apps layer
- **ARD** — client-side discovery of external MCP servers and A2A agents at runtime

## Model providers and tool bridges

LLM access is its own surface: configure a provider once, then manage the models agents may use. See [Providers](/platform/providers).

External tools and context arrive through **MCP servers**, attached as capabilities so an agent gains them without changes to the runtime.

## Evaluation

- **Braintrust** — connect observability and evaluation workflows to agent traces and real runs

## Adding integrations

New integrations follow a documented parity checklist before release, so each one lands with a consistent surface across the API, SDKs, and operator console.

## Related

- [Providers](/platform/providers) — connecting Everruns to LLM vendors
- [Sandboxes](/platform/sandboxes) — isolated execution environments
- [Integrations reference](https://docs.everruns.com/integrations/) — full setup details for each integration