8 views
-/https://github.com/berriai/litellm/issues/30126
GitHub · issue

#30126 [Bug]: `/cursor/chat/completions` is not logged in LiteLLM 1.88.1

  • State: open
  • Author: @edwinyoo44
  • Labels: bug, proxy

### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

## `/cursor/chat/completions` returns 200 and provider is charged, but no LiteLLM SpendLogs / Logs UI entry is created

I configured Cursor IDE according to the official LiteLLM Cursor integration docs:

https://docs.litellm.ai/docs/tutorials/cursor_integration

Cursor Base URL:

```text https://example.com/cursor ````

With this setup, Cursor sends requests to:

```text POST /cursor/chat/completions ```

The request succeeds and LiteLLM returns HTTP 200. The upstream provider also shows usage/cost, but LiteLLM Logs UI / SpendLogs does not show any corresponding record.

Example LiteLLM access log:

```text INFO 52.44.113.131:55762 - "POST /cursor/chat/completions HTTP/1.1" 200 | timestamp=2026-06-10T21:15:20.332082 component=uvicorn.access logger=h11_impl.py:473 ```

A/B test:

* `https://example.com/v1` → `POST /v1/chat/completions` → LiteLLM logs are created normally * `https://example.com/cursor` → `POST /cursor/chat/completions` → provider is charged, but no LiteLLM log is created

LiteLLM version:

```text 1.88.1 ```

Expected behavior:

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (4 events)
#0 of 0 · 31d18h52m54s ago — entered · #import:https:::github.com:berriai:litellm post #2001
The left issue likely requires tracing a specialized proxy route through request handling, callbacks, persistence, and spend tracking, with regression coverage across integrations. The right issue is primarily a targeted pricing-data correction with validation, making it substantially smaller in scope and risk.
19499 spans asynchronous execution, algorithmic performance, hook inheritance, and security-sensitive behavior, requiring broader investigation and regression testing. 30126 is more likely a localized proxy-route observability integration fix.
The right issue is harder because it spans a nonstandard proxy route, request lifecycle, provider usage accounting, logging persistence, and regression coverage, whereas the left issue appears localized to model-list response metadata and formatting. The broader cross-cutting behavior and risk of affecting billing and observability justify a 3:1 effort ratio.
#0 of 0 · 31d17h41m34s ago — current · #import:https:::github.com:berriai:litellm post #3196
31510 requires security-sensitive guardrail coverage across multiple request schemas, shared extraction logic, and broader regression testing; 30126 is more localized request-path/logging integration work.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search