Hey, what’s the best way to use MCP with Framer?

I use Cursor daily and recently started experimenting with the Framer MCP through the Claude extension in Cursor.

One thing I’m finding a bit clunky is having to remind Claude every session which Framer project I’m working on and what the project context is.

For those running multiple Framer projects, what’s worked best for you?

A separate MCP/database per Framer project with all project context stored there?

One master knowledge base with branches/namespaces per project?

Project-specific CLAUDE.md / rules files and a shared MCP backend?

Something else entirely?

Curious how others are handling project memory, context management, and switching between projects without constantly re-explaining things to the agent.

Hey, what’s the best way to use MCP with Framer?

I use Cursor daily and recently started experimenting with the Framer MCP through the Claude extension in Cursor.

One thing I’m finding a bit clunky is having to remind Claude every session which Framer project I’m working on and what the project context is.

For those running multiple Framer projects, what’s worked best for you?

A separate MCP/database per Framer project with all project context stored there?

One master knowledge base with branches/namespaces per project?

Project-specific CLAUDE.md / rules files and a shared MCP backend?

Something else entirely?

Curious how others are handling project memory, context management, and switching between projects without constantly re-explaining things to the agent.