Problemsby IbisTrouble with Claude anthropic
LLM apps8 posts from 8 people7d active+22 posts in the last 7 days, 1 the 7 days before (steady)
Posts per day
The posts behind it
8, newest first[BUG] Claude Sonnet 5.5 / Opus 5.5 on native Anthropic provider will return 400 when one tool is passed with available_functions### Description With exactly one tool and using the available_functions parameter, AnthropicCompletion._prepare_completion_params sets tool_choice: {"type": "tool"}. This was fine for older models such as Claude Sonnet 5 and below but Claude Sonnet 5.5, Opus 5.5, Fable 5.1 and...crewAIInc/crewAI · darrentirtoOpen ↗AI Agent: Anthropic thinking block is duplicated across parallel tool calls, Anthropic API rejects the next turn with 400 "Invalid `signature` in `thinking` block"...+ `find-users`, `get-user` + `get-invoices`, `find-users` + `find-teams`, …). In every failing run, both actions in `subNodeExecutionData.actions[].action.metadata.anthropic` had the identical `thinkingSignature`. ### Expected behavior Parallel Anthropic tool calls from a sinn8n-io/n8n · josiasmontag1 repliesOpen ↗anthropic: treat Bedrock Claude model IDs like their canonical names in the mid-conversation system message check### Submission checklist - [x] This is a feature request, not a bug report or usage question. - [x] I added a clear and descriptive title that summarizes the feature request. - [x] I used the GitHub search to find a similar feature request and didn't find it. - [x] I checked the...langchain-ai/langchain · AlcebiadesFilho2 repliesOpen ↗[BUG] AI Gateway silently drops Anthropic `cache_control` on the normalized chat route (prompt caching unreachable)> [!WARNING] > Before submitting a PR, please make sure that: > - A maintainer has triaged this issue and applied the `ready` label > - This issue has no assignee > - No duplicate PR exists > > PRs not meeting these requirements may be automatically closed. ### Issues Policy...mlflow/mlflow · Nantina2 repliesOpen ↗AI Agent: Anthropic thinking replay sends earlier tool calls back as `input: {}`...5 runs, and 0 of 8 on an earlier `claude-sonnet-4-6` run (no thinking by default). Workaround: set the Anthropic Chat Model node's Thinking Mode to Disabled. ### To Reproduce I haven't built a stand-alone minimal workflow. These steps come from the code and my own workflow. 1....n8n-io/n8n · CurioBeyond1 repliesOpen ↗[BUG] Anthropic model prefix filtering rejects custom deployments (refs #5893)## Description `LLM._matches_provider_pattern` (`lib/crewai/src/crewai/llm.py:405`) only accepts `["claude-", "anthropic."]` for the Anthropic provider, while the module constant `ANTHROPIC_PREFIXES` is `("anthropic/", "claude-", "claude/")` - the two lists are inconsistent, and...crewAIInc/crewAI · kpdagrt22Open ↗Python: [Bug]: Anthropic connector silently drops ImageContent in user messages and crashes on image-only requests### Describe the bug In `semantic-kernel` Python SDK, `_format_user_message` in `semantic_kernel/connectors/ai/anthropic/services/utils.py` currently hardcodes: This causes two major issues when interacting with multimodal Anthropic Claude models (such as Claude 3.5 Sonnet,...microsoft/semantic-kernel · chauvuusvn1 repliesOpen ↗ProviderToolSearchMiddleware rejects ChatAnthropicBedrock (provider 'anthropic_bedrock' missing from the server tool search registry)...only has `anthropic` and `openai`. So `_prepare_request` raises: ### Reproduction ### Workaround / suggested fix Map `anthropic_bedrock` to the Anthropic descriptor, since it is the same server tool: Once that works, a second issue shows up: `langchain_anthropic` sends the...langchain-ai/langchain · adirAfek1 repliesOpen ↗
Companies and products named
About this problem
Evidence
8 posts from 8 people in 5 places, about 3 a week over 19 days. Mostly on langchain-ai/langchain, n8n-io/n8n, crewAIInc/crewAI. Tools named alongside: OpenAI, LangChain, Amazon Bedrock, Gemini.
Frustration Frustration 0 of 3· Seen on GitHub issues
How it was grouped
Posts about Claude with no known issue, sharing the word "anthropic". First post Sep 28, 2026, latest Oct 16, 2026. Corroborated: 3 or more posts from 2 or more people or places. Method
Other Claude problems
Claude app gets 1 and 2 star reviewsClaude rate limits and usage caps block workClaude runs out of context on real workClaude loses context in large codebasesClaude upgrades break working setupsTrouble with Claude desktopClaude fails to return valid structured outputClaude login, SSO or OAuth breaksClaude lock-in makes switching costlyClaude writes buggy code that is hard to reviewTrouble with Claude agentTrouble with Claude content
History
fading to steady
steady to fading
rising to steady
steady to rising
rising to steady
unverified to corroborated
fading to rising
Trouble with Claude anthropic (1 post)