Grab it from "Try it yourself" — it carries the three things any agent needs to operate a tool.
Connect MCPs. Build in one place
MCP lets an AI agent reach into the tools you already use and do the work — not just talk about it.
Story behind
My work lives across a dozen tools.
Spreadsheets, Notion, Slack, Airtable — and none of them talk to each other.
So I became the glue.
Copy a row from a sheet into Notion.
Paste a summary into Slack.
Re-type the same thing in three places.
The AI could tell me what to do,
but it couldn't touch any of my tools.
So what if it could reach into the tools I already use,
and do the work itself?
That's what MCP does.
It's a standard plug that lets an agent connect to a tool and operate it.
You give the agent three things:
- Access — an MCP connection or API key
- How to build — the tool's docs and rules
- What to build — the outcome you want, in plain words
Now "summarise these rows and post them to Slack" isn't an afternoon.
It's one sentence to my agent.
Why this helps you
How to
- Connect your tool over MCP or an API key, so the agent can act — not just advise.
- Skip if you're unsure — the AI will tell you what to connect and share.
Tell it, in plain words, what you're trying to do.
- Share your goal and how you work.
- Let the AI recommend which connectors to add and what context to give.
Run one small action first to confirm it can actually reach the tool.
Let it do the real task end to end, then check the result in the tool.
Good to know
Copy the skill
Build your first thing with MCP
Paste it to your agent as the first message.
More how-tos
Let AI build your Airtable
I let an AI agent build Airtable bases through the API, and flag what I must finish by hand.
Ship your own website with an AI code agent
I left Framer's €24/month to build my own site with an AI agent — full design control, near-zero cost.
Build a second brain for your project work
I stopped re-explaining myself to every AI by giving it one folder it reads first — my context, on tap.