Timed out requests

Support Bug Reports
1

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

Request ID: 2cbad8b6-f2d9-4562-bb31-0a6d52ed2833
[internal] Failed to run step, exceeded max retries
aae: [internal] Failed to run step, exceeded max retries
at Pnw (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:43963:24479)
at Bnw (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:43963:23385)
at $nw (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:43964:6487)
at v9u.run (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:43964:11286)
at async JIn.resume (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:56307:15131)
at async C0d.streamFromAgentBackend (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:56377:11057)
at async C0d.getAgentStreamResponse (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:56377:17161)
at async ZOe.submitChatMaybeAbortCurrent (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:44075:19892)

Steps to Reproduce

Its a simple request i am asking and i am getting this error. If i open other driectories, i dont have this issue, only seems to be in the current repo i am working in. Seems to always stick between "planning next moves, summarizing chat and reconnecting before i get this error.

Expected Behavior

I normally get a response.

Operating System

Windows 10/11

Version Information

Latest version of cursor

For AI issues: which model did you use?

Sonnet, Composer 2. I have tried various models.

For AI issues: add Request ID with privacy disabled

Request ID: 2cbad8b6-f2d9-4562-bb31-0a6d52ed2833
[internal] Failed to run step, exceeded max retries
aae: [internal] Failed to run step, exceeded max retries
at Pnw (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:43963:24479)
at Bnw (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:43963:23385)
at $nw (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:43964:6487)
at v9u.run (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:43964:11286)
at async JIn.resume (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:56307:15131)
at async C0d.streamFromAgentBackend (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:56377:11057)
at async C0d.getAgentStreamResponse (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:56377:17161)
at async ZOe.submitChatMaybeAbortCurrent (vscode-file://vscode-app/c:/Users/josep/AppData/Local/Programs/cursor/resources/app/out/vs/workbench/workbench.desktop.main.js:44075:19892)

Does this stop you from using Cursor

No - Cursor works, but with this issue

5

Hey, thanks for the report with the Request ID. It helped us find the cause quickly.

The issue is that the context from this repo is over the model token limit 200k in non-max mode. Cursor tries to shrink the context with summarization, but it is not enough, so it keeps retrying until it runs out of retries. That is why it works in other projects where the context is smaller.

Workarounds:

  • Start a new chat instead of continuing a long one, long sessions build up context
  • Try MAX mode, it supports a higher token limit
  • Split tasks into smaller requests so the agent does not pull in too much context at once
  • If you are using Composer 2, try switching to a specific model like Claude Sonnet 4.6 or GPT-5

The team is aware of this issue. The problem is that summarization cannot recover after going over the limit. There is no ETA for a fix yet, but your report helps us prioritize it.

Let me know if these workarounds helped.