Integrations

Why integrations matter

Enrai is not only a browser-based coding workspace. It is also designed to be driven by external systems:

• workflow tools
• agent orchestrators
• internal platforms
• CI-like automation entrypoints

That is why the task queue exists behind a simple HTTP API and supports rollout continuation, task callbacks, and agent-friendly execution semantics.

Example: n8n

One practical setup is using n8n as the orchestration layer:

1. A webhook in n8n receives an event from your internal system.
2. n8n prepares the task payload and calls POST /api/enrai/tasks.
3. Enrai runs the coding task in sequence.
4. When the task finishes, Enrai calls your callback_url.
5. n8n receives that callback and continues the flow:
   • notify Slack
   • create a ticket comment
   • trigger a deployment review
   • enqueue another task

Typical payload sent from n8n:

{
  "text": "Add a health endpoint and run tests",
  "workdir": "/workspace/apps/api",
  "callback_url": "https://automation.example.com/enrai/task-finished"
}

Example: OpenClaw or other agent orchestrators

If you already have a higher-level agent runtime such as OpenClaw, Enrai can be used as the coding execution surface behind that system.

In that model:

• the orchestrator decides what to work on
• Enrai performs the coding task against a real workspace
• the resulting rollout becomes a durable artifact
• the callback or queue state feeds status back into the orchestrator

That setup is useful when you want:

• a dedicated coding environment
• rollout history per task
• terminal and file access
• a human-reviewable workspace after the agent finishes

Design note

Enrai is intentionally structured so it can be handled by agents as well as by human operators.

That shows up in four places:

• tasks can be created through HTTP
• sessions can continue existing rollouts
• callbacks can notify other systems when work ends
• the editor still remains available for human inspection and follow-up

Recommended pattern

Use Enrai as the execution layer, not the whole orchestration layer.

That usually means:

• n8n, OpenClaw, or another workflow engine decides when to run a task
• Enrai executes that task safely and serially
• another system consumes the result and decides what happens next