chore(lint): fix post-merge type regressions and ratchet lint budgets

basedpyright-code-budget.json

@@ -1,9 +1,9 @@
 {
   "reportAny": {
-    "limit": 17270
+    "limit": 16279
   },
   "reportArgumentType": {
-    "limit": 2539
+    "limit": 2530
   },
   "reportAssignmentType": {
     "limit": 319
@@ -24,7 +24,7 @@
     "limit": 19
   },
   "reportExplicitAny": {
-    "limit": 5486
+    "limit": 5063
   },
   "reportFunctionMemberAccess": {
     "limit": 7
@@ -42,7 +42,7 @@
     "limit": 12
   },
   "reportIndexIssue": {
-    "limit": 35
+    "limit": 30
   },
   "reportInvalidTypeForm": {
     "limit": 34
@@ -54,10 +54,10 @@
     "limit": 0
   },
   "reportMissingParameterType": {
-    "limit": 5658
+    "limit": 5642
   },
   "reportMissingTypeArgument": {
-    "limit": 15425
+    "limit": 15404
   },
   "reportMissingTypeStubs": {
     "limit": 40
@@ -105,19 +105,19 @@
     "limit": 109
   },
   "reportUnknownMemberType": {
-    "limit": 38721
+    "limit": 38622
   },
   "reportUnknownParameterType": {
-    "limit": 19778
+    "limit": 19748
   },
   "reportUnknownVariableType": {
-    "limit": 30290
+    "limit": 30210
   },
   "reportUnnecessaryCast": {
     "limit": 117
   },
   "reportUnnecessaryComparison": {
-    "limit": 697
+    "limit": 696
   },
   "reportUnnecessaryContains": {
     "limit": 5

litellm/caching/valkey_semantic_cache.py

@@ -19,7 +19,7 @@ import hashlib
 import os
 from collections.abc import Mapping, Sequence
 from dataclasses import dataclass
-from typing import Any, Final, Protocol
+from typing import Any, Final, Protocol, cast
 
 from redis import Redis
 from redis.asyncio import Redis as AsyncRedis
@@ -294,7 +294,8 @@ class ValkeySemanticCache(RedisSemanticCache):
                 print_verbose("No prompt provided for semantic caching")
                 return
 
-            embedding: Final = await self._get_async_embedding(prompt, metadata=kwargs.get("metadata"))
+            metadata: Final = cast("dict[str, object] | None", kwargs.get("metadata"))  # cast-ok: untyped kwargs
+            embedding: Final = await self._get_async_embedding(prompt, metadata=metadata)
             await self._ensure_index_async(len(embedding))
 
             doc_key: Final = self._doc_key(key)

litellm/integrations/arize/arize_phoenix_prompt_manager.py

@@ -4,7 +4,7 @@ Fetches prompt versions from Arize Phoenix and provides workspace-based access c
 """
 
 from collections.abc import Mapping, Sequence
-from typing import Any, Final
+from typing import Any, Final, cast
 
 from jinja2 import DictLoader, select_autoescape
 from jinja2.sandbox import ImmutableSandboxedEnvironment
@@ -203,7 +203,9 @@ class ArizePhoenixTemplateManager:
             # Combine rendered content
             final_content = " ".join(rendered_content_parts)
 
-            rendered_messages.append({"role": role, "content": final_content})
+            rendered_messages.append(
+                cast("AllMessageValues", {"role": role, "content": final_content})  # cast-ok: Phoenix roles are OpenAI
+            )
 
         return rendered_messages
 

litellm/llms/anthropic/experimental_pass_through/context_management/editors/compact.py

@@ -36,6 +36,7 @@ from litellm.types.llms.anthropic import (
     CompactionBlock,
     UsageIteration,
 )
+from litellm.types.llms.openai import AllMessageValues
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.streaming_handler import CustomStreamWrapper
@@ -867,7 +868,7 @@ def _append_text_to_content(content: object, extra_text: str) -> object:
 
 
 class _SummaryCallUserKwarg(TypedDict, total=False):
-    user: ReadOnly[object]
+    user: ReadOnly[str]
 
 
 class _SummaryCallRegionKwarg(TypedDict, total=False):
@@ -876,11 +877,11 @@ class _SummaryCallRegionKwarg(TypedDict, total=False):
 
 class _SummaryCallKwargs(TypedDict):
     model: ReadOnly[str]
-    messages: ReadOnly[list[dict[str, object]]]
+    messages: ReadOnly[list[AllMessageValues]]
     max_tokens: ReadOnly[int]
     timeout: ReadOnly[float]
     litellm_metadata: ReadOnly[Mapping[str, object]]
-    user: NotRequired[ReadOnly[object]]
+    user: NotRequired[ReadOnly[str]]
     allowed_model_region: NotRequired[ReadOnly[str]]
 
 
@@ -927,11 +928,15 @@ async def _call_summary_model(
     end_user_id: Final = metadata.get("user_api_key_end_user_id")
     call_kwargs: Final[_SummaryCallKwargs] = {
         "model": summary_model,
-        "messages": summary_messages,
+        "messages": cast("list[AllMessageValues]", summary_messages),  # cast-ok: built as OpenAI chat messages
         "max_tokens": max_tokens,
         "timeout": COMPACT_SUMMARY_TIMEOUT_SECONDS,
         "litellm_metadata": metadata,
-        **(_SummaryCallUserKwarg(user=end_user_id) if end_user_id else _SummaryCallUserKwarg()),
+        **(
+            _SummaryCallUserKwarg(user=end_user_id)
+            if isinstance(end_user_id, str) and end_user_id
+            else _SummaryCallUserKwarg()
+        ),
         **(
             _SummaryCallRegionKwarg(allowed_model_region=allowed_model_region)
             if allowed_model_region is not None

litellm/llms/vertex_ai/common_utils.py

@@ -1,7 +1,7 @@
 import re
 from copy import deepcopy
 from enum import Enum
-from typing import Any, Final, Literal, get_type_hints
+from typing import Any, Final, Literal, cast, get_type_hints
 
 import httpx
 
@@ -726,8 +726,9 @@ def set_schema_property_ordering(schema: dict[str, object], depth: int = 0) -> d
             schema["propertyOrdering"] = [k for k, v in schema["properties"].items()]
         for k, v in schema["properties"].items():
             set_schema_property_ordering(v, depth + 1)
-    if "items" in schema:
-        set_schema_property_ordering(schema["items"], depth + 1)
+    items: Final = schema.get("items")
+    if isinstance(items, dict):
+        set_schema_property_ordering(cast("dict[str, object]", items), depth + 1)  # cast-ok: JSON Schema child
     return schema
 
 

litellm/proxy/guardrails/guardrail_hooks/bedrock_guardrails.py

@@ -364,9 +364,7 @@ class BedrockGuardrail(CustomGuardrail, BaseAWSLLM):
         bedrock_request["content"] = bedrock_request_content
         return bedrock_request
 
-    def _build_response_content_items(
-        self, response: object, has_grounding: bool
-    ) -> list[BedrockContentItem]:
+    def _build_response_content_items(self, response: object, has_grounding: bool) -> list[BedrockContentItem]:
         """Build content item(s) from the model response. When the request supplied
         grounding, the response is qualified ``guard_content`` so Bedrock can score it.
         """

litellm/proxy/litellm_pre_call_utils.py

@@ -7,7 +7,7 @@ from collections import OrderedDict
 from collections.abc import Mapping, MutableMapping, Sequence
 from datetime import datetime
 from types import MappingProxyType
-from typing import TYPE_CHECKING, Any, Final
+from typing import TYPE_CHECKING, Any, Final, cast
 
 from fastapi import HTTPException, Request
 from pydantic import ValidationError as PydanticValidationError
@@ -1343,6 +1343,8 @@ class LiteLLMProxyRequestSetup:
     def get_sanitized_user_information_from_key(
         user_api_key_dict: UserAPIKeyAuth,
     ) -> StandardLoggingUserAPIKeyMetadata:
+        stripped_metadata: Final = strip_callback_config(user_api_key_dict.metadata)
+        auth_metadata: Final = cast("dict[str, str] | None", stripped_metadata)  # cast-ok: metadata is free-form JSON
         user_api_key_logged_metadata: Final = StandardLoggingUserAPIKeyMetadata(
             user_api_key_hash=user_api_key_dict.api_key,  # just the hashed token
             user_api_key_alias=user_api_key_dict.key_alias,
@@ -1365,7 +1367,7 @@ class LiteLLMProxyRequestSetup:
             user_api_key_budget_reset_at=(
                 user_api_key_dict.budget_reset_at.isoformat() if user_api_key_dict.budget_reset_at else None
             ),
-            user_api_key_auth_metadata=strip_callback_config(user_api_key_dict.metadata),
+            user_api_key_auth_metadata=auth_metadata,
         )
         return user_api_key_logged_metadata
 

litellm/proxy/management_endpoints/key_management_endpoints.py

@@ -2476,6 +2476,9 @@ async def _validate_update_key_data(
     user_api_key_cache: UserApiKeyCache,
 ) -> None:
     """Validate permissions and constraints for key update."""
+    if prisma_client is None:
+        raise HTTPException(status_code=500, detail={"error": "Database not connected"})
+
     # Reject NaN/±inf spend before it can reach the DB / spend counter.
     validate_finite_spend(data.spend)
     validate_budget_duration(data.budget_duration)
@@ -2594,7 +2597,7 @@ async def _validate_update_key_data(
     # _check_key_admin_access that would otherwise require team/org admin status.
     _key_is_team_key: Final = getattr(existing_key_row, "team_id", None) is not None
     can_skip_admin_check: Final = (caller_is_creator or _key_is_team_key) and not _is_budget_change
-    if (not _is_proxy_admin) and prisma_client is not None and not can_skip_admin_check:
+    if (not _is_proxy_admin) and not can_skip_admin_check:
         hashed_key: Final = existing_key_row.token
         await _check_key_admin_access(
             user_api_key_dict=user_api_key_dict,

ruff-strict-budget.json

@@ -1,21 +1,21 @@
 {
   "ANN001": {
-    "limit": 3012
+    "limit": 3004
   },
   "ANN002": {
     "limit": 71
   },
   "ANN003": {
-    "limit": 827
+    "limit": 825
   },
   "ANN201": {
     "limit": 2003
   },
   "ANN202": {
-    "limit": 845
+    "limit": 843
   },
   "ANN204": {
-    "limit": 702
+    "limit": 700
   },
   "ANN205": {
     "limit": 112
@@ -24,7 +24,7 @@
     "limit": 133
   },
   "ANN401": {
-    "limit": 655
+    "limit": 517
   },
   "ASYNC230": {
     "limit": 11
@@ -168,7 +168,7 @@
     "limit": 3
   },
   "RET504": {
-    "limit": 175
+    "limit": 174
   },
   "RUF012": {
     "limit": 239
@@ -198,7 +198,7 @@
     "limit": 58
   },
   "SIM102": {
-    "limit": 315
+    "limit": 313
   },
   "SIM103": {
     "limit": 119
@@ -231,7 +231,7 @@
     "limit": 5
   },
   "TID251": {
-    "limit": 1117
+    "limit": 1105
   },
   "TRY002": {
     "limit": 524

type-discipline-budget.json

@@ -1,9 +1,9 @@
 {
   "LIT001": {
-    "limit": 22705
+    "limit": 22655
   },
   "LIT002": {
-    "limit": 26854
+    "limit": 26830
   },
   "LIT003": {
     "limit": 269
@@ -27,10 +27,10 @@
     "limit": 0
   },
   "LIT010": {
-    "limit": 16564
+    "limit": 16546
   },
   "LIT011": {
-    "limit": 5577
+    "limit": 5558
   },
   "LIT012": {
     "limit": 4508