litellm - 💡(How to fix) Fix [Bug]: Claude Code 2.1.104 throws Extra inputs are not permitted [2 comments, 1 participants]

{ "status": "failure", "max_retries": null, "batch_models": null, "usage_object": null, "user_api_key": "<REDACTED>", "cost_breakdown": null, "attempted_retries": null, "error_information": { "traceback": " File "/usr/lib/python3.13/site-packages/litellm/proxy/anthropic_endpoints/endpoints.py", line 53, in anthropic_response\n result = await base_llm_response_processor.base_process_llm_request(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n ...<16 lines>...\n )\n ^\n File "/usr/lib/python3.13/site-packages/litellm/proxy/common_request_processing.py", line 1065, in base_process_llm_request\n responses = await llm_responses\n ^^^^^^^^^^^^^^^^^^^\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 5144, in async_wrapper\n return await self._ageneric_api_call_with_fallbacks(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n ...<2 lines>...\n )\n ^\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 3800, in _ageneric_api_call_with_fallbacks\n raise e\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 3787, in _ageneric_api_call_with_fallbacks\n response = await self.async_function_with_fallbacks(**kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 5613, in async_function_with_fallbacks\n return await self.async_function_with_fallbacks_common_utils(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n ...<8 lines>...\n )\n ^\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 5570, in async_function_with_fallbacks_common_utils\n raise original_exception\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 5528, in async_function_with_fallbacks_common_utils\n raise original_exception\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 5604, in async_function_with_fallbacks\n response = await self.async_function_with_retries(*args, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 5759, in async_function_with_retries\n self.should_retry_this_error(\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^\n error=e,\n ^^^^^^^^\n ...<4 lines>...\n content_policy_fallbacks=content_policy_fallbacks,\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n )\n ^\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 5963, in should_retry_this_error\n raise error\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 5710, in async_function_with_retries\n response = await self.make_call(original_function, *args, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 5878, in make_call\n response = await response\n ^^^^^^^^^^^^^^\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 3927, in _ageneric_api_call_with_fallbacks_helper\n raise e\n File "/usr/lib/python3.13/site-packages/litellm/router.py", line 3913, in _ageneric_api_call_with_fallbacks_helper\n response = await response # type: ignore\n ^^^^^^^^^^^^^^\n File "/usr/lib/python3.13/site-packages/litellm/utils.py", line 2093, in wrapper_async\n raise e\n File "/usr/lib/python3.13/site-packages/litellm/utils.py", line 1892, in wrapper_async\n result = await original_function(*args, **kwargs)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File "/usr/lib/python3.13/site-packages/litellm/llms/anthropic/experimental_pass_through/messages/handler.py", line 268, in anthropic_messages\n response = await init_response\n ^^^^^^^^^^^^^^^^^^^\n File "/usr/lib/python3.13/site-packages/litellm/llms/custom_httpx/llm_http_handler.py", line 1968, in async_anthropic_messages_handler\n raise self._handle_error(\n ~~~~~~~~~~~~~~~~~~^\n e=e, provider_config=anthropic_messages_provider_config\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n )\n ^\n File "/usr/lib/python3.13/site-packages/litellm/llms/custom_httpx/llm_http_handler.py", line 4721, in _handle_error\n raise provider_config.get_error_class(\n ...<3 lines>...\n )\n", "error_code": "400", "error_class": "BaseLLMException", "llm_provider": "", "error_message": "{"message":"context_management: Extra inputs are not permitted"}" }, "applied_guardrails": null, "user_api_key_alias": "<REDACTED>", "spend_logs_metadata": null, "user_api_key_org_id": null, "proxy_server_request": null, "requester_ip_address": null, "user_api_key_team_id": "<REDACTED>", "user_api_key_user_id": "<REDACTED>", "guardrail_information": [ { "duration": 0.000344, "end_time": 1776855325.671803, "risk_score": 0, "start_time": 1776855325.67146, "guardrail_id": "<REDACTED>", "match_details": null, "guardrail_mode": "pre_call", "guardrail_name": "secret-filter", "policy_template": null, "detection_method": null, "guardrail_status": "success", "patterns_checked": 6, "guardrail_provider": "litellm_content_filter", "guardrail_response": [], "masked_entity_count": {} } ], "model_map_information": null, "mcp_tool_call_metadata": null, "additional_usage_values": {}, "cold_storage_object_key": null, "user_api_key_project_id": null, "user_api_key_team_alias": "<REDACTED>", "litellm_overhead_time_ms": null, "user_api_key_project_alias": null, "vector_store_request_metadata": null }

TL;DR

Downgrade claude-code to version 2.1.98 to resolve the HTTP 400 error caused by extra inputs not being permitted.

Guidance

• The error message {"message":"context_management: Extra inputs are not permitted"} suggests that the issue is related to the claude-code version 2.1.104 not handling extra inputs correctly.
• Downgrading to 2.1.98 has been confirmed to fix the issue, indicating a potential regression in the newer version.
• Review the litellm configuration and guardrail settings to ensure that no extra inputs are being passed to the claude-code API.
• Verify that the ANTHROPIC_BASE_URL, ANTHROPIC_AUTH_TOKEN, and other environment variables are set correctly and not causing any issues.

Notes

• The root cause of the issue is likely a change in the claude-code API that is not compatible with the current litellm version v1.83.7.
• Further investigation is needed to determine the exact cause of the issue and to find a long-term solution that does not require downgrading claude-code.

Recommendation

• Apply workaround: Downgrade claude-code to version 2.1.98 until a fix is available for the newer version.