Error listings
Developers with new AMD GPUs (Radeon RX 9070) experience segfaults when Ollama partially offloads model inference to GPU via ggml backend tensor operations Includes evidence for Ollama troubleshooting demand.
partial offload segfaults in ggml_backend_tensor_set on AMD Radeon RX 9070 Developers using the new mcp_toolset beta feature encounter intermittent parsing errors when model returns generic tool_use events instead of expected mcp_tool_use events Includes evidence for Anthropic API troubleshooting demand.
model intermittently emits tool_use instead of mcp_tool_use for tools declared in mcp_toolset Developers integrating Ollama via Python API get token-too-long errors when llama-server outputs responses exceeding bufio.Scanner buffer limits, breaking LLM pipeline processing Includes evidence for Ollama troubleshooting demand.
bufio.Scanner: token too long (status code: -1) — Error from the python API: llama-server response Fix CodeInterpreter error where reasoning-type items must be paired with content items in the OpenAI Responses API, avoiding validation failures in multi-step reasoning workflows Includes evidence for OpenAI API troubleshooting demand.
Error: Item 'rs_ABCD' of type 'reasoning' was provided without its required... Fix error when specifying explicit tool_choice (e.g., 'required', 'none', or function name) with GPT-5 models, forcing fallback to auto mode Includes evidence for OpenAI API troubleshooting demand.
Tool choices other than 'auto' are not supported with model on GPT5 Fix Vercel deployment failures caused by macOS/Windows case-insensitive filesystem masking import path casing mismatches that fail on Vercel's Linux environment Includes evidence for Vercel troubleshooting demand.
Module not found — Can't resolve './Components/Header' (import path casing mismatch) Fix SDK hang bug where responses.create() blocks forever with GPT-5 models, while identical requests succeed via raw REST API — requiring migration away from Python SDK Includes evidence for OpenAI API troubleshooting demand.
responses.create() hangs indefinitely with GPT-5 models; direct REST API call completes successfully Fix sudden 401 authentication failure when upgrading azure-openai SDK that previously worked on older versions Includes evidence for OpenAI API troubleshooting demand.
AAD bearer token passed via api_key works in 2.33.0 but returns 401 in 2.34.0 and after Resolve network connectivity error that completely blocks paid Claude Code users from accessing AI features Includes evidence for Claude Code troubleshooting demand.
ConnectionRefused issue. Can't use AI at all on a PAID plan Fix Microsoft 365 MCP integration that stopped working in recent Claude Code release Includes evidence for Claude Code troubleshooting demand.
MCP connector to M365 Not Working Fix self-hosted runner not receiving any jobs despite being properly configured with correct repository and group permissions Includes evidence for GitHub Actions troubleshooting demand.
Org-level self-hosted runner never dispatched despite correct runner group repo access Get proactive notifications via alerts or webhooks before Anthropic API session or weekly budget caps are reached, enabling proactive scaling or budget management Includes evidence for Anthropic API troubleshooting demand.
No built-in mechanism to receive alerts or webhooks before session/weekly budget limits are exhausted — users discover rate limits only after hitting 429 Fix Claude Code authentication routing so Max plan users are billed under their personal subscription, not an Organization API key Includes evidence for Claude Code troubleshooting demand.
Claude Code OAuth authenticates successfully but binds to Organization's API billing instead of personal Max subscription; /status shows 'API Usage Billing' instead of 'Max Plan' Understand whether ChatGPT Plus subscription quota applies to third-party OAuth tool calls and how to resolve 429 quota errors when using third-party tools like OpenClaw Includes evidence for OpenAI API troubleshooting demand.
ChatGPT Plus OAuth succeeds but third-party Codex calls return 429 quota exceeded — Plus subscription does not provide usable quota for third-party tool integration Understand why usage spikes right after rate limit reset and how to avoid being unexpectedly throttled despite hitting reset Includes evidence for Claude Code troubleshooting demand.
Rate limit reset calculation incorrect — usage spike detected immediately after rate limit resets, suggesting burst behavior or race condition in token accounting Prevent ultraReview from consuming free quota credits when run fails mid-pipeline and ensure failed runs report a distinct error status instead of falsely reporting completed+[] Includes evidence for Claude Code troubleshooting demand.
/ultrareview fails at Verify stage (Review failed), Dedupe never starts, yet both failed runs count against free quota (3 free runs); completion reports status: completed with empty findings array [] making failures indistinguishable from clean reviews Resolve spurious session limit errors on Pro plan where user has not exceeded any visible usage thresholds Includes evidence for Claude Code troubleshooting demand.
You've hit your session limit · resets at scheduled time — Pro plan user unable to continue working despite not actively using the service Fix 401 Unauthorized error when using Azure AD bearer tokens passed via api_key parameter after upgrading to openai Python SDK 2.34.0+, specifically in Azure APIM proxy setups Includes evidence for OpenAI API troubleshooting demand.
Error code: 401 - {'error': {'code': '401', 'message': 'Access denied due to invalid subscription key or wrong API endpoint. Make sure to provide a valid key for an active subscription and use a correct regional API endpoint for your resource.'}} Fix intermittent AttributeError crash in AsyncAnthropicBedrock streaming when first SSE event is a rate_limit_error; users cannot access event.message.model or event.message.usage because message is None Includes evidence for Anthropic API troubleshooting demand.
AttributeError: 'NoneType' object has no attribute 'model'; AttributeError: 'NoneType' object has no attribute 'usage' — triggered by Bedrock SSE error payload {"type": "error", "error": {"type": "rate_limit_error", "message": "Rate limited"}} Fix LiteLLM proxy TPM rate limiting in distributed deployments where each replica independently enforces the limit, effectively multiplying the allowed throughput beyond intended cap Includes evidence for LiteLLM troubleshooting demand.
TPM enforcement allows tpm_limit × N_replica concurrent requests across replicas — single pod limit fails to apply correctly in multi-pod K8s deployments