MCP Server¶
Summary¶
Model Context Protocol (MCP) is an open standard for connecting AI assistants to external tools, data sources, and systems. MCP servers provide the "hands" — the ability for agents to interact with the outside world.
How It Works¶
MCP servers expose tools and resources that AI agents can discover and call. The agent decides which tools to use based on the task at hand.
Role in Agent Architecture¶
| Component | Role | Analogy |
|---|---|---|
| Model | Processing power | CPU |
| Agent runtime | Orchestrates context | OS |
| MCP servers | Connection to outside world | Network/IO |
| Skills | Expertise and procedural knowledge | Applications |
"MCP is the hands. Skills are the experience." — Barry Zhang
MCP + Skills¶
Developers are building skills that orchestrate workflows of multiple MCP tools stitched together. MCP provides the connection to external data; skills provide the expertise for what to do with it.
Examples in the Wild¶
- Tolgee — First localization platform with native MCP server at
/mcp/developer; lets AI coding assistants manage translations directly (search keys, create translations, trigger MT) - Notion — MCP skills that help Claude understand workspaces and do deep research
- Browserbase — MCP skill for Stagehand browser automation
- GitHub — MCP tools for code search, PR management, issue creation
See Also¶
- Agent Skills
- Agentic Workflows
tolgee