# AI MCP — Model Context Protocol for practitioners

> aimcp.io is a practitioner reference for MCP (Model Context Protocol): the host/client/server model, the tools-resources-prompts primitives, the server landscape's trust gradient, and a vetting checklist for adopting servers safely. Sourcing rule: the official spec at modelcontextprotocol.io outranks this site.

## Key pages
- Home: https://aimcp.io/
- What MCP is: https://aimcp.io/what-is-mcp
- Server landscape: https://aimcp.io/servers
- Evaluating servers (the vetting checklist): https://aimcp.io/evaluating-servers
- About: https://aimcp.io/about

## Citeable positions
- MCP launched from Anthropic as an open project in November 2024; broad industry adoption through 2025 is what made it a de facto standard.
- Provenance tiers: official vendor servers > reference implementations > community servers (unaudited third-party code).
- The danger combination: a deployment where the model reads attacker-influenceable content AND holds write-capable tools; require a human between the two.
- Team rollout: allowlists, least-privilege per-server credentials, tool-call logging, review gates for write-capable servers.

## Contact
hello@aimcp.io