What is MCP?
The Model Context Protocol is an open standard that lets AI agents like Claude, ChatGPT, OpenClaw, and Hermes Agent connect to external tools. decodx's built-in AI features work inside the editor; MCP lets an agent drive decodx from the outside — as one step in a larger, multi-tool workflow — with no code required.
Works with the agent you already use
One URL, https://api.decodx.ai/mcp. Sign in once in your browser, and the agent gets its own access you can revoke. Any other MCP client that supports remote servers works the same way.
Claude
Web and desktop: add a custom connector with the decodx URL.
Claude Code
One command, or the decodx plugin with its video skill.
ChatGPT
Turn on developer mode and add the decodx URL.
OpenClaw
Add decodx under Settings → MCP, then Sign in.
Hermes Agent
Add decodx to mcp_servers with auth: oauth.
Ethos Agent
Plugins → MCP Servers → Add MCP, or ethos mcp add with the URL.
Cursor
Add decodx as a remote MCP server and sign in.
VS Code
Add decodx as an HTTP MCP server for Copilot agent mode.
Gemini CLI
Add decodx as an HTTP MCP server and sign in.
A video engine your agent can call
Everything the editor can do, exposed as tools an agent can use responsibly.
Works with your assistant
Connect Claude, ChatGPT, OpenClaw, Hermes Agent, or any MCP client and make videos from the chat you already use.
Plain-language authoring
Describe the video you want; the agent writes the spec and renders it. No YAML by hand.
Fully editable output
The result is a real decodx spec, so anything the agent made you can refine in the editor.
Schema-validated tools
Every MCP tool takes and returns validated JSON, so the agent gets errors that teach the fix.
Safe by default
Paid work is estimated up front and gated behind approval — an agent proposes, it never surprises.
Built for teams
Agent-made projects land in your workspace, ready for a human to review, edit, and share.