PR #25449: fix(prompt_management): return passthrough when prompt_id is None in sync path

• Repository: BerriAI/litellm
• Author: jeanibarz
• State: closed | merged: False
• Link: https://github.com/BerriAI/litellm/pull/25449

Description (problem / solution / changelog)

Relevant issues

Fixes #25425

What's the problem?

When prompt_id is None, the sync get_chat_completion_prompt() in PromptManagementBase raises a hard ValueError:

ValueError: prompt_id is required for Prompt Management Base class

This causes 500 errors when prompt management hooks are triggered without a prompt_id — for example, when vector_store_ids: [] is silently injected into model settings via the UI (see #25133), causing should_run_prompt_management_hooks() to return True.

The async counterpart async_get_chat_completion_prompt() handles this correctly by delegating to should_run_prompt_management(), which returns False for None prompt_id in all implementations (dotprompt, langfuse, generic, gitlab, etc.), causing a graceful passthrough.

Before (sync path — crashes):

if prompt_id is None:
    raise ValueError("prompt_id is required for Prompt Management Base class")

After (sync path — passthrough, matching async):

# Removed — should_run_prompt_management() already handles None prompt_id
if not self.should_run_prompt_management(...):
    return model, messages, non_default_params

Verification

Tested with a concrete PromptManagementBase subclass where should_run_prompt_management() returns False for None prompt_id (same as dotprompt, langfuse, generic, gitlab implementations):

• Old code: ValueError: prompt_id is required for Prompt Management Base class
• New code: returns (model, messages, non_default_params) unchanged — matching async behavior

Changes

• Removed the ValueError guard in get_chat_completion_prompt() so the sync path uses the same should_run_prompt_management() check as the async path
• Added 3 unit tests covering: sync None passthrough, async None passthrough, and sync with a valid prompt_id

Pre-Submission checklist

• I have Added testing in the tests/test_litellm/ directory, Adding at least 1 test is a hard requirement - see details
• My PR passes all unit tests on make test-unit
• My PR's scope is as isolated as possible, it only solves 1 specific problem
• I have requested a Greptile review by commenting @greptileai and received a Confidence Score of at least 4/5 before requesting a maintainer review

Type

🐛 Bug Fix

Changed files

• litellm/integrations/prompt_management_base.py (modified, +0/-2)
• tests/test_litellm/integrations/test_prompt_management_base.py (added, +162/-0)