I’m running Claude Code within Trae AI IDE, and so far it’s working well for actual development work. The MCP support has been great, though there are some limitations I’m working around.
The Setup and Agent Selection
I can select specific agents for specific projects, which is genuinely useful. The ability to have different AI personalities or capabilities for different types of work makes a difference in output quality.
The MCP integration means I can connect to various services directly from the IDE, which streamlines a lot of workflows that would otherwise require switching between tools.
Current Limitations: Multiple MCPs
The one thing I can’t figure out yet is configuring multiple MCPs for multiple projects. I’d love to have:
- Multiple Postgres connections for different projects
- Multiple Cloudflare account connections
- Multiple GitHub repo configurations
Right now it’s a little annoying to manage, but it’s not a deal breaker.
Real Usage and Costs
I tried the pro plan for 3 days, and now I’m on the max 100 plan. Based on my stats, I’m averaging about $60 per day. I don’t think I need the max 200 plan yet, but it’s still too early to tell.
I’ve been running two projects simultaneously:
- Website building - Frontend development and deployment
- Postgres SQL queries - Database analysis and optimization
Claude Code vs Chat Interface
This is really different from just using the blog desktop or chat interface. I’m using Trae AI’s Sonnet 4 through their subscription for regular chat, but it’s just not as effective.
The chat version is okay, but it doesn’t fix everything the way Claude Code does. Claude Code is different—it actually understands context better and provides more actionable solutions for development work.
The Development Experience
Running actual development projects through this setup has been surprisingly smooth. The ability to have the AI understand project structure, make file changes, and run commands all within the same environment removes a lot of friction.
For website building, having instant feedback and iteration capabilities changes how quickly you can prototype and refine ideas.
For database work, being able to write, test, and optimize queries with AI assistance in real-time is genuinely valuable.
Cost vs Value Reality
$60/day sounds expensive until you consider what it’s replacing:
- Faster development cycles
- Fewer context switches between tools
- Real-time problem solving
- Multiple project management
If this is saving me several hours per day on development work, the cost becomes justifiable pretty quickly.
What’s Next
I need to keep experimenting with the MCP configurations to see if I can solve the multiple-project connection issue. The workflow is solid enough that I’m willing to work around current limitations.
The combination of Claude Code’s capabilities with Trae AI’s IDE integration is promising for anyone doing serious development work, especially if you’re managing multiple projects with different service dependencies.
Still early days, but the $60/day investment is looking worthwhile for actual productivity gains.