fix(rerank): log optional_rerank_params at debug to stop leaking request content

[devin-ai-integration]

Relevant issues

Fixes #32525

Linear ticket

Pre-Submission checklist

• I have added meaningful tests
• My PR passes all CI/CD checks (e.g., lint, format, unit tests)
• My PR's scope is as isolated as possible; it only solves 1 specific problem
• I have received a Greptile Confidence Score of at least 4/5 before requesting a maintainer review

Screenshots / Proof of Fix

Live proxy at LITELLM_LOG=INFO routing to real AWS Bedrock (bedrock/amazon.rerank-v1:0), same curl in both runs:

curl -s http://localhost:PORT/rerank -H "Content-Type: application/json" -H "Authorization: Bearer sk-1234" -d '{
  "model": "rerank-model",
  "query": "MARKER_QUERY_sensitive_user_search",
  "documents": ["MARKER_DOC_confidential_customer_text", "unrelated filler doc"],
  "top_n": 2
}'

Before (line at verbose_logger.info, the current main behavior); grep optional_rerank_params on the proxy log shows the raw request content at INFO:

19:13:02 - LiteLLM:INFO: main.py:161 - optional_rerank_params: {'query': 'MARKER_QUERY_sensitive_user_search', 'documents': ['MARKER_DOC_confidential_customer_text', 'unrelated filler doc'], 'top_n': 2, 'rank_fields': None, 'return_documents': None, 'max_chunks_per_doc': None}

After (this PR, commit 24df024); same curl, same INFO log level, grep optional_rerank_params and grep MARKER_ on the proxy log return nothing:

=== INFO log lines leaking content (after fix) ===
0
0

Both runs reached real AWS Bedrock (the account here lacks the bedrock:Rerank IAM action so the upstream call returns 403), which is fine because the leak happens at optional_rerank_params logging before the provider call; the point is whether that content lands in logs at the default INFO level

Type

🐛 Bug Fix

Changes

litellm/rerank_api/main.py logged the mapped optional_rerank_params at INFO. That dict always carries the caller's query and documents, so every rerank call wrote raw request content to stdout and any log sink at the default LITELLM_LOG=INFO. Because it is a plain verbose_logger call outside the logging object, it also bypasses turn_off_message_logging and the redact_messages guards. This was the only completion/embedding/rerank path emitting request content at INFO; sibling code (search/main.py, ocr/main.py, litellm_core_utils/litellm_logging.py) already logs optional_params at debug

- verbose_logger.info(f"optional_rerank_params: {optional_rerank_params}")
+ verbose_logger.debug(f"optional_rerank_params: {optional_rerank_params}")

Added a regression test at tests/test_litellm/rerank_api/test_main.py that drives litellm.arerank (cohere provider, mocked transport) with marker query/documents and asserts no INFO-or-above LiteLLM log record contains the request content, while the optional_rerank_params line is still emitted at DEBUG. It fails on the pre-fix verbose_logger.info and passes on the fix

Link to Devin session: https://app.devin.ai/sessions/67a91a96b957427895a7f3aff032c5da

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[greptile-apps]

Greptile Summary

This PR fixes a log-level issue in litellm/rerank_api/main.py where optional_rerank_params — which always contains the caller's query and documents — was emitted at INFO, making raw request content visible in default log output and any downstream log sink. The fix is a single-line change from verbose_logger.info to verbose_logger.debug, consistent with how sibling paths (search/main.py, ocr/main.py) already handle the same log call.

• litellm/rerank_api/main.py: verbose_logger.info → verbose_logger.debug for the optional_rerank_params line; no logic change.
• tests/test_litellm/rerank_api/test_main.py: New regression test that drives litellm.arerank with a mocked HTTP transport, asserts marker content does not appear in any INFO-or-above record, and confirms the log line is still emitted at DEBUG.

Confidence Score: 5/5

Safe to merge — the change is a single log-level demotion with no effect on request handling or response behavior.

The only modified production line changes a log call from INFO to DEBUG; all request-handling logic is untouched. The new test mocks the HTTP transport (no real network calls), directly exercises the changed line, and would fail against the pre-fix code. No backwards-incompatible changes or auth-path modifications.

No files require special attention.

Important Files Changed

Filename	Overview
litellm/rerank_api/main.py	Single-line change: verbose_logger.info → verbose_logger.debug for optional_rerank_params, matching the pattern already used by search/main.py and ocr/main.py
tests/test_litellm/rerank_api/test_main.py	New regression test using mocked HTTP transport (no real network calls); asserts marker content absent from INFO+ records and present at DEBUG, correctly targeting the "LiteLLM" logger name
tests/test_litellm/rerank_api/init.py	Empty init.py added to make the new test directory a Python package

_{Reviews (1): Last reviewed commit: "fix(rerank): log optional_rerank_params ..." | Re-trigger Greptile}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[codspeed-hq]

Merging this PR will not alter performance

✅ 30 untouched benchmarks

---

_{Comparing litellm_fix_rerank_info_log_32525 (32a08f9) with litellm_internal_staging (b00877c)}