A place for developers to talk about building on Farcaster.

/dev

What's really is interesting is that there is an outsized usage in one of our MCP tools: postgres. There are people using a crazy amount of postgres MCP calls with skeet.build

2hrs ago we didn’t have MySQL and now we do

🗽based; 

building skeet.build 🟠;
building @buoy 🛟; 
building check.supply 🏦;

@reachbot - experimenting with AI & Frames | Custom Keyword Alerts in TG t.me/CastAlertsBot | Human-readable EVM data https://github.com/3loop/loop-decoder

Yea we’re in the middle of writing docs for each mcp and recommendations.

How to setup:

1. You need a staging or a dev MySQL or Postgres db in the cloud. We say dev because primarily it’s for development.

2. The main benefits will be that cursor can query for your table schemas. So for example you need ti build like a model will all the typing it can do that in one shot. Or you need to build a new api endpoint with a new table, it can create the table, create the endpoint for you. The tool that the mcp server and cursor is using is basically the shell of the db so it can literally execute any query you want it to.

3.  You should set it so that the user has read only and dev db thats recoverable to start until you feel comfortable. This is recommended because cursor potentially could execute a query that nukes your db like drop or flush. We highly do not recommend prod unless it’s read only.

> each schema in folders

I don’t think the organization of your code matters to cursor. You would just @ mention the file or folder.

You’re doing this to pull context into the cursor composer so that it can more accurately do what you tell it to do.

If there are tables for example you’re updating the schema or need to update your code our mcp server would crush it. Or audit it.

How does integration with a db work? I'm using Cursor and usually a monorepo structure with all schemas in folders. Curious to understand if there are any benefits to trying out mpc