Skip to main content
Pleyor ships a Model Context Protocol server that lets you build, run, and manage agents directly from any MCP-compatible chat client. Add it once as a connector and your assistant can search your agents, build new workflows from a description, trigger runs, wait for results, and show the live canvas — all without leaving the chat.

​
What you can do once it’s connected

Build workflows from a description

“Build me a workflow that takes a brand image and writes three promotional captions.” Your assistant designs the graph, validates it, saves it, and shows the canvas inline.

Trigger runs and wait for results

“Run my podcast-summary agent on this episode.” The connector handles inputs, blocks until terminal, and surfaces the asset URLs.

See the canvas in chat

The same studio-fidelity canvas you use in the Pleyor studio renders directly in the chat with live per-node status — no tab-switching needed.

Iterate and recover

Runs failed? The assistant reads the typed error code, follows the recovery playbook (retry, change model, adjust input), and re-runs.

​
Get started

Quickstart

Add the connector to Claude or ChatGPT in two minutes.

Tool reference

The full surface — what each tool does and when to call it.

In-chat canvas

How the studio renders inside the chat, what’s interactive, what isn’t.

Authentication

OAuth for chat clients, API keys for scripts and server-to-server.

​
What MCP is, in one line

MCP is an open standard that lets an LLM call your server’s tools and read your server’s resources. The chat client (Claude, ChatGPT, etc.) is the “host”; Pleyor’s server is one of many possible “servers” the host can connect to. Once you add Pleyor as a connector, the model sees Pleyor’s tools next to its built-in ones and can call them as part of a conversation. The protocol itself: modelcontextprotocol.io.