Understanding Connectors
Connectors are the core building blocks of Findexar. Understanding how they work is essential for setting up your ChatGPT apps and agents.
What Are Connectors?
Connectors are MCP (Model Context Protocol) servers that enable communication between ChatGPT and your backend services. Findexar uses two types of connectors:
- Downstream Connectors - Your backend MCP servers
- Upstream Connectors - Findexar's MCP endpoints for ChatGPT
Downstream Connectors
Downstream connectors are the backend services that do the actual work. They can be:
Supported Types
- Zapier MCP Server - Connect to Zapier automations
- Make MCP Server - Connect to Make.com workflows
- n8n MCP Server - Connect to n8n automations
- OpenAI App SDK MCP - Custom MCP servers built with OpenAI's SDK
- Custom MCP Servers - Any MCP-compatible service
- Findexar Service - Built-in service for minimal setup (useCredits tool)
How They Work
Downstream connectors:
- Expose tools that can be called by ChatGPT
- Execute the actual business logic
- Return results to Findexar
- Can be hosted anywhere (your servers, cloud, etc.)
Upstream Connectors
Upstream connectors are Findexar's MCP endpoints that ChatGPT connects to. They:
- Handle authentication and user context
- Manage billing and usage tracking
- Route requests to downstream connectors
- Provide tools to ChatGPT
Creating an Upstream Connector
When you create an upstream connector:
- Select which downstream connectors to use
- Choose which tools to expose
- Set credit prices for each tool
- Configure authentication and access
The Flow
Here's how requests flow through the system:
- End User interacts with your ChatGPT app or agent
- ChatGPT calls a tool from your Upstream Connector (Findexar)
- Findexar authenticates the user and checks credits
- Findexar routes the request to the appropriate Downstream Connector
- Downstream Connector executes the work and returns results
- Findexar deducts credits and logs usage
- Results flow back to ChatGPT and the end user
Tool Mapping
Tool mapping is the process of:
- Selecting tools from downstream connectors
- Exposing them through your upstream connector
- Setting credit prices for each tool
- Configuring tool behavior and access