What Is MCP?
Learn what Model Context Protocol is and how it connects AI agents to tools, resources, prompts, and APIs.
Quick Answer / TL;DR
Model Context Protocol is a standard way for AI clients to discover and call external tools, read resources, and use prompt templates without custom integration code for every app.
Key Takeaways
- MCP separates AI clients from data providers.
- Tools perform actions while resources provide read-only context.
- Stdio is common locally; remote HTTP streaming is better for teams.
Core architecture
An MCP client such as Claude Desktop, Cursor, or a custom agent opens a connection to one or more MCP servers. The server declares capabilities through machine-readable schemas, and the client chooses which tools or resources to expose to the model.
This keeps the model interface consistent even when the backend is a database, file system, CRM, payment API, government dataset, or internal application.
| Primitive | Purpose | Example |
|---|---|---|
| Tool | Execute an approved action | Create a GitHub issue |
| Resource | Read context | Load a Postgres schema |
| Prompt | Reuse instructions | Run a code review prompt |
India-first implementation note
Indian teams often combine SaaS connectors with local business systems such as UPI, GST, logistics, education, and healthcare datasets. MCP works well when those systems already expose APIs but need a safer model-facing orchestration layer.
{
"mcpServers": {
"india-edge-api": {
"url": "https://mcpserver.in/v1/mcp",
"headers": {
"Authorization": "Bearer ${MCP_API_KEY}",
"X-Data-Region": "in"
}
}
}
}What Is MCP? FAQs
Direct answers for developers, operators, and Indian teams evaluating MCP.