Agentic Workflows — Mastery2 / 8
MCP Servers 101 — Give Your Agent Real Tools
The Model Context Protocol is how your agent stops guessing and starts querying your database, your issues, your browser. Here is the mental model and the first connection.
An agent with no tools is a very articulate guesser. MCP — the Model Context Protocol — is the open standard that turns guessing into querying. Plug in a server and your agent can read your database, open your issues, or drive a browser directly.
The mental model
MCP has two sides: a server exposes capabilities; a client (Claude Code, here) uses them. A server can expose three things:
- Tools — actions the agent can call (run a query, create an issue).
- Resources — data the agent can read, referenced with
@. - Prompts — pre-built prompts that show up as slash commands.
Three transports
| Transport | Use it for |
|---|---|
| stdio | Local servers run as a subprocess — files, git, local scripts. |
| SSE | Remote servers over HTTP streaming. |
| HTTP | Remote servers over streamable HTTP. |
Your first connection
claude mcp add --transport http linear https://mcp.linear.app/mcp
Then run /mcp inside Claude Code to authenticate and confirm the connection.
Trust is the whole game
Once a server is connected, its tools appear to the agent automatically (named mcp__<server>__<tool>) and you approve them under /permissions like any other tool. Next: building your own server, so the agent can reach the one system nobody has an off-the-shelf MCP for — yours.
Series — Agentic Workflows — Mastery
- Part 01The Agentic Workflow MindsetMost developers treat AI like a smarter autocomplete. The ones who pull ahead treat it like a teammate that acts. Here is the mental model.
- Part 02MCP Servers 101 — Give Your Agent Real Tools — you are hereThe Model Context Protocol is how your agent stops guessing and starts querying your database, your issues, your browser. Here is the mental model and the first connection.
- Part 03Build Your First Custom MCP ServerOff-the-shelf servers cover GitHub and Postgres. The high-leverage one is the server only you can write — the bridge to your own system.
- Part 04Hooks — Make the Agent Obey Your RulesA prompt asks the model to remember. A hook makes it happen — deterministically, every time, outside the model's control.
- Part 05Custom Slash Commands as Team WorkflowsA custom slash command is a reusable prompt you commit to the repo — so the whole team runs the same high-quality instruction instead of re-typing it.
- Part 06Subagents — Delegating Work That ScalesOne giant context gets slow and vague. Subagents let the main agent hand focused work to specialists with their own context and tools — and run them in parallel.
- Part 07The Daily-Driver Setup — Settings, Permissions, Status LineThe difference between fighting the agent and flowing with it is twenty minutes of configuration you do once. Here is the setup.
- Part 08Multi-Tool Pipelines — Ticket to Reviewed BranchThe payoff: chain MCP, hooks, commands and subagents into one flow that takes a ticket to a reviewed branch — with you in the loop only where it counts.