Skip to content
Docs

API & Automation

Interplay is automatable end to end: the flows you build become APIs, and the platform that runs them is itself driven by an API — so deployment and flow promotion can be scripted into an existing CI/CD pipeline.

How do you turn a flow into an API?

Section titled “How do you turn a flow into an API?”

Add an HTTP trigger node to a flow, wire the logic, and add an HTTP response node. On deploy, the flow is live at its configured path and can be called by any client. The same flow can serve multiple routes, and pipelines group flows behind a coordinated set of routes for larger multi-stage services.

What else can trigger a flow?

Section titled “What else can trigger a flow?”
  • Schedules — cron-style timing for recurring jobs.
  • Inbound events — webhooks, MQTT, TCP/UDP, and WebSocket messages.
  • Streams — file watches, directory drops, and mail streams.
  • Another flow — flows can invoke each other, so shared logic is written once.
  • Chat — a chatbot widget or an agent conversation can drive a flow.
  • Manual runs — from the editor, for testing and one-off operations.

Is there an administrative API?

Section titled “Is there an administrative API?”

Yes. Interplay exposes an authenticated administrative API that covers the operations the editor performs — deploying and updating flows, listing and searching flows, managing projects and pipelines, installing and removing nodes, managing Python environments and packages, configuring agents and MCP servers, workspace file operations, and querying version history. Authentication is by API key or administrative credentials.

How do you automate deployment in CI/CD?

Section titled “How do you automate deployment in CI/CD?”

The common pattern is:

  1. Treat flow JSON as source, reviewed and versioned in Git like any other artifact.
  2. Build an image or package per environment, or keep one image and vary environment variables.
  3. On release, call the administrative API to push the new flow definitions to the target environment.
  4. Reload and redeploy flows in place — Interplay can reload flows from storage and deploy them without restarting the process, so routine changes ship with no downtime.
  5. Verify with the health endpoint and the flow’s own HTTP routes.

Can Interplay be called by other AI agents?

Section titled “Can Interplay be called by other AI agents?”

Yes. Interplay can expose capabilities as an MCP server, which lets external agents, IDEs, and assistants discover and invoke Interplay flows as tools. Combined with HTTP triggers, this means a flow built once is reachable from a browser, a backend service, and an AI agent without being rebuilt for each.

Sources and references

Section titled “Sources and references”