MCP Integration - Token size issue

[onestreamlive]

Description

I am using mcp server at https://github.com/GLips/Figma-Context-MCP for figma design importing.

My model has 80k context window set. When i ask kilo tool to fetch figma design , it answers with following errors.

The MCP tool executed successfully, but the output is unavailable, because it is too long (657311 estimated tokens, limit is 80000 tokens). If you need the output, find an alternative way to get it in manageable chunks.

If estimated is 657311 tokens and limit is 80000 tokens, then how come it cannot fit in that context?

Any one??

[chrarnoldus]

If estimated is 657311 tokens and limit is 80000 tokens, then how come it cannot fit in that context?

657 311 is a lot more than 80 000, that's why it doesn't fit

[xoraingroup]

Oh sorry totally missed it. Do i need to increase context window of model for that?

[chrarnoldus]

What model are you using? Some models support a 1M context but I'm not sure if it will really work that well.

[xoraingroup]

I am using qwen-coder3

[Dave-London]

Structured output from MCP tools is one way to prevent this. I built
https://github.com/Dave-London/pare — a suite of MCP servers that wrap dev CLIs and return compact JSON via structuredContent instead of raw terminal output.

In benchmarks, verbose tool output (build logs, test results, install progress) sees
80-95% token reduction. For already-compact output, the savings are smaller or neutral. But for the kind of large outputs that hit context limits, the reduction is dramatic.

[johnnyeric]

Thanks for taking the time to report this!

We've just released a new version of the Kilo Code extension. As part of this transition, we're automatically closing issues that were filed against the legacy extension.

If you're still experiencing this issue with the new extension, feel free to reopen this or open a fresh issue with updated details - we'd love to hear from you. We're actively working to make Kilo Code better, and your feedback is a big part of that.

Thanks again for being part of the community!