Build an MCP Server
The MCP ecosystem is the extension model for AI Chat.
If Skills package strategy behavior, MCP servers package tool behavior.
What MCP Is Used For
In Orien AI, MCP servers are used to:
- extend AI Chat with external tools
- provide reusable data and action capabilities
- create a marketplace around chat-side operator tooling
Current Product Model
The backend already supports a marketplace flow for MCP servers:
- browse approved servers
- view server detail
- install and uninstall
- enable and disable installed servers
- review and rate servers
- publish new servers
Server Metadata
When publishing an MCP server, the platform expects metadata such as:
slugnamedescriptioncategorytransport- declared
tools - optional
required_config
Supported transports currently include:
stdiossestreamable-http
Why MCP Matters
MCP is part of the answer to a missing trading agent ecosystem.
It gives AI Chat a real extension surface:
- tools can be reused
- developers can publish and distribute capabilities
- operators can assemble better chat workflows without rebuilding infrastructure
Relationship to Skills
MCP and Skills are parallel ecosystems:
- MCP extends AI Chat
- Skills extend pipeline and integrated agents
They should share the same product center, but they should not collapse into one system.