"Message Too Long" error on all queries

Bug Reports
1

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

Since updating to the latest version, started getting “Message Too Long” errors on all queries, no matter how short or what model is used.

My current query is literally 122 characters right now and does not call any files.

Steps to Reproduce

Send any query!

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

Version: 2.1.17 (Universal)
VSCode Version: 1.105.1
Commit: 6757269838ae9ac4caaa2be13f396fdfbcf1f9a0
Date: 2025-11-21T08:55:17.983Z (13 hrs ago)
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Darwin arm64 24.3.0

For AI issues: which model did you use?

All models (Sonnet 4.5/Sonnet 4.5 Thinking, Grok Code, Gemini 3 Pro, etc) return the same error.

For AI issues: add Request ID with privacy disabled

d857b3ad-d59f-42e3-9b6f-318c28d81bb4

Does this stop you from using Cursor

Yes - Cursor is unusable

2

Hey, thanks for the report. This is unusual, because the error appears even with very short requests.

Could you please help us gather some diagnostic information:

  • Screenshot: Please share a screenshot of the actual error message.
  • .cursorrules file: Check if you have a .cursorrules file in the root of your project and, if so, what its size is.
  • New workspace test: Does this happen in a completely new/empty workspace, or only in your current project?
  • MCP servers: Are your MCP servers enabled? Each MCP server contributes to context, and if you have many servers configured, this could be part of the issue.
  • Large context: Are you adding folders/files with a very large number of lines to the context?
  • Codebase indexing: Check the codebase indexing status.

In the meantime, you can try:

  • Temporarily disabling codebase indexing to see if context calculation is causing the issue.
  • Testing in a new empty folder to check if the problem is specific to this project.

With this additional information, I can escalate the issue to engineering if needed, since it looks like this might be a regression in version 2.1.17.