How to overcome token limits?

[will-feetham]

I am struggling to use this MCP in Claude Code in Cursor because even with simple workflows (>20 nodes) the read function exceeds token limits:

● Read(......\Downloads+Write First Resume Per Job.json)

⎿ Error: File content (48232 tokens) exceeds maximum allowed tokens (25000). Please use offset
and limit parameters to read specific portions of the file, or use the GrepTool to search for
specific content.

How are people overcoming this challenge? Love the tool, just need to make it work as well as it does for >10 node workflows!

[jayc-codes]

I had similar issues with claude desktop and was concerned about cursor and windsurf having limits too. I thought the answer was to use openwebui (no limits) and deepseek (no limits and cheap) #46
I managed to get it working but its not as slick as claude.

We can use any kind of llm (local/external) with openwebui and mcpo as a tool without any limits, we just need to know how to best configure openwebui with mcpo? @czlonkowski any tips please?