Both sit in the tools & integrations layer with similar commercial terms, so the practical split is language: MCP (Model Context Protocol) targets Python and Arcade targets Python.
The open protocol for exposing tools and data to agents. The default answer.
Write your integrations as MCP servers and every harness on this page can use them unchanged.
Tool-calling with per-user OAuth, so the agent acts as the user, not as root.
The per-user token model is what stops one user's agent reading another user's mailbox.
| MCP (Model Context Protocol) | Arcade | |
|---|---|---|
| Licence | Open source | Open source |
| Pricing | open source | freemium |
| Self-hostable | Yes | Yes |
| Languages | TypeScript, Python | Python, TypeScript |
| Install | npm install @modelcontextprotocol/sdk | — |
| Keys required | None | ARCADE_API_KEY |
Both sit in the tools & integrations layer with similar commercial terms, so the practical split is language: MCP (Model Context Protocol) targets Python and Arcade targets Python. MCP (Model Context Protocol): The open protocol for exposing tools and data to agents. The default answer. Arcade: Tool-calling with per-user OAuth, so the agent acts as the user, not as root.
MCP (Model Context Protocol) can run on your own infrastructure. Arcade can run on your own infrastructure.
MCP (Model Context Protocol) is open source (open source). Arcade is open source (freemium).
Every page here answers to Accept: text/markdown and returns the same content at roughly a tenth the tokens. No separate site, no toggle — same URL.
curl -s -H "Accept: text/markdown" https://newagent.build/compare/mcp-vs-arcade