XIO

xWire

Posts from every feed QUEST has found across the web — blogs, news, podcasts, channels and more.

Cursor - Community Forum - Topics tagged anthropic

Topics tagged anthropic

Anthropic Guardrails (Opus 5.5 / Fable 5.2)

Hello, folks, I am running into this issue when trying to use Fable 5.2 or Opus 5.5. The request is pretty run of the mill. I can’t tell where in the chain (context, skills, MCPs) this goes wrong. I hope you can help. Request ID: 5dfba1b9-ffe0-47a2-8467-7e96dc6e0b98 {“error”:“ERROR_OPENAI”,“details”:{“title”:“Request blocked by Anthropic”,“detail”:“We are unable to complete this request because it was blocked under Anthropic’s Usage Policy.”,“isRetryable”:…

Cloud Desktop inaccessible — Claude Sonnet usage limit despite Grok 4.6 selected

Describe the Bug Hello Cursor Support Team, I need assistance with a persistent Cloud Desktop issue that is blocking my website development. I use Cursor Cloud Agents exclusively through Chrome, without the desktop application. Problem: The Desktop panel displays: “The agent’s machine may not be running. Send a message to restart.” The agent reports that its cloud machine is running and the website is available internally, but the interactive Desktop cannot connect. I am using Grok 4.6 High…

Dancing Cat: a cat that dances while your AI is thinking 🐱

I made a small extension: a cat dances while Claude Code is thinking and naps when it’s done. It also dances for Cursor’s own agent, because Cursor runs Claude Code hooks. Shows in a panel, floating next to your cursor, or both Use your own cat GIFs Free, open source, nothing leaves your computer Install: search “Dancing Cat for Claude Code” in Extensions, or get it on Open VSX Setup: after installing, click Set Up in the Dancing Cat panel. GitHub: Fateme-mokhtari/claude-dancing-cat It’s my…

Claude Code chat import silently drops sessions when project path contains spaces

Where does the bug appear (feature/product)? Cursor IDE Describe the Bug Cursor’s Claude Code importer builds project folder names by only replacing slashes with dashes. Claude Code replaces every non-alphanumeric character with a dash. For a project at /Users/USER/Local Sites/SITE/PROJECT, Claude Code stores sessions under ~/.claude/projects/-Users-USER-Local-Sites-SITE-…, while Cursor looks for ~/.claude/projects/-Users-USER-Local Sites-SITE-… I created the exact symlink Cursor expects. It…

Ability to pin skill to model

Feature request for product/service Cursor IDE Describe the request The ability to pin a skill to a model would allow for cost-savings without the fuss of manually selecting a model before executing the skill. Claude supports this, but Cursor does not. Cursor supports a subagent pinned to a model, but the delegation costs negate much of the savings and, in some cases, cost more as seen below. What we measured We tried each workaround Cursor offers today on two of our skills. We used the same…

Waiting for good non-anthropic models on cursor

I understand Anthropic’s rapid development of new models each month. However, a significant portion of users desire models like Kimi 3.1 Pro and Muse Ember. The wait for Fable 5.2’s release and availability on the cursor is minimal, just an hour, compared to weeks and days for others. While I may not be fully aware of the internal workings of the cursor, I can confidently say it’s not currently the best option for users. A focus on expanding the range of available models would be greatly…

Making Cursor Cloud Agents even better

Feature request for product/service Cloud Agent (GitHub, Slack, Linear,…) Describe the request Our whole team uses Cursor Cloud Agents for our AI-driven SDLC - kicking off tasks, getting PRs back, running work in parallel. It’s become a core part of how we ship, and we love it. At the same time, we have a significant amount of Anthropic API credits that we can’t use with Cloud Agents. Today our Anthropic key works for chat in the editor, but Cloud Agents ignore it and always bill through Cursor…

Fable 5.2 BYOK errors + News on Azure Foundry?

Bedrock us.anthropic.claude-fable-5-2: “The model returned the following errors: `temperature` is deprecated for this model.” Can we all… generally agree that Bedrock *.anthropic.claude-fable-5* and *.anthropic.claude-opus-5* no longer require temperature? Why can’t we just implement this as “don’t pass the temperature argument for fable >= 5 and opus >= 5.5 and sonnet >= 5.5” Also… any update on getting Foundry integrated? It should be trivial… 6 posts - 5 participants Read full topic

Cursor Pro Usage - Opus burned 100% in 1.5 messages

I know these kind of messages come up a lot, so I tried to avoid them or “being that guy”… Additionally, I also make a point to attempt to be very Context/Token efficient, using Context-Mode, JCodeMunch, CodeGraph, LLMTrim, and one or two more other tools - all in an attempt to be more cost effective with my models, Cursors or otherwise. Ontop of that, I had not used my Cursor subscription for 2+ months, paying it, but having been out of the country for family matters for most of it, among a…

AWS Bedrock: Model name is not valid for all explicit Bedrock IDs (Auto works)

Where does the bug appear (feature/product)? Cursor IDE Describe the Bug AWS Bedrock is enabled and credentials appear to work (catalog returns Bedrock models; Auto succeeds with BYOK routing). But selecting any explicit Bedrock model ID fails immediately with:AI Model Not Found Model name is not valid: “” Steps to Reproduce Open Cursor Settings → Models → enable AWS Bedrock, enter Access Key / Secret / Region. Confirm models refresh succeeds (Bedrock models appear in the picker; locally we see…

Anthropic BYOK still unusable while "Override OpenAI Base URL" is set — one global toggle breaks Claude BYOK (Windows, 3.27.22)

Where does the bug appear (feature/product)? Cursor IDE Describe the Bug Anthropic BYOK only works when “Override OpenAI Base URL” is OFF. With the Override ON (needed for an OpenAI-compatible gateway/proxy), selecting any Claude model with my own Anthropic key fails every time: ERROR_BAD_USER_API_KEY / “Invalid API key. Unauthorized User API key” Request ID: 718e39b2-d9b3-42fd-8ee2-af741ef16721 The key is valid. After turning the Override OFF, the same key works and Anthropic bills the…

Pro+ account: Network Error / resource_exhausted whenever I select GPT or Claude models

Where does the bug appear (feature/product)? Cursor IDE Describe the Bug Hi Cursor team, I’m on the Pro+ plan, but I’m unable to use other models. Whenever I select a different GPT or Claude model and submit a prompt, Cursor displays a network error. Retrying the request produces the same error. I have also tried selecting several different GPT and Claude models, but all of them fail in the same way. My usage page shows 0% usage, so this does not appear to be caused by reaching my usage limit.…

Most suitable model for generating documentation

I am using Cursor to generate product documentation particularly user guide, release notes, manuals from sources that I provide. To be able to generate almost close to manually written documents, any suggestion on which is the best model? I have tried Composer 2.5, Claude Opus 5.5 so far. Claude Opus has given good results. But any other tried and tested suggestions would be great to know. Of course, something budget-friendly is a must. 3 posts - 4 participants Read full topic

"AI Model Not Found" error when running "/summarize"

Where does the bug appear (feature/product)? Cursor IDE Describe the Bug When attempting to run “/summarize” command while using Claude Sonnet 5.5, I get the following error… AI Model Not Found: Invalid parameters for registry model: “claude-sonnet-5-5” Seems to be an issue with the way 5.5 is being handled / expected. Steps to Reproduce Start a chat with Claude Sonnet 5.5 model selected. Attempt to run /summarize to condense context window. Expected Behavior Expect chat to summarize and…

AWS Bedrock unknown_parameter

Where does the bug appear (feature/product)? Cursor IDE Describe the Bug I am using AWS Bedrock and added Access Key ID and Secret Access Key in Cursor. I am using Claude Opus 5.5 Bedrock (US) 300K on Bedrock. I also tried Claude Opus 4.6 Bedrock (US) 200K and it gives the same error. It is now affecting all other models. When I used Claude Opus 5.5 through Cursor also same error. If I use Grok 4.7 it returns “This model does not support custom API keys.” No models can be used. I set AWS…

Enterprise BYOK + AI Gateway: built-ins ignore Override Base URL; need Anthropic override

Feature request for product/service Cursor IDE Describe the request Feature request for product/service Cursor IDE — Bring Your Own Key / Override OpenAI Base URL (Enterprise) Describe the request I’m on Cursor Enterprise (Trilogy Cursor; same email as my Cursor account). I use Cursor alongside Claude and OpenAI. Separately, I want to route as much local Chat/Agent inference as possible through our TrueFoundry AI Gateway (Maat PAT + Override OpenAI Base URL to the company cursor/v1 gateway…

Sensitive Prompt Error when using Opus or Sonet models

Sonet or Opus models are not accepting any prompts, responding with below error. Agent encountered an error We are unable to complete this request because it triggered restrictions on cyber content and was blocked under [Anthropic’s Usage Policy]( Cyber Verification Program | Claude Help Center ). Try a less sensitive prompt. 4 posts - 2 participants Read full topic

AWS Bedrock (BYOK): chats fail with "Invalid `signature` in `thinking` block ... bound to a different conversation

Describe the Bug Title: AWS Bedrock (BYOK): chats fail with “Invalid signature in thinking block … bound to a different conversation” Summary When using Claude through AWS Bedrock with my own keys, chats work for a while, then fail mid-conversation with a thinking-block signature error. Starting a new chat fixes it temporarily, until it happens again. Error The model returned the following errors: messages.1.content.7: Invalid signature in thinking block. The block is bound to a different…

[Bug] All models fail with resource_exhausted on new conversations - Pro+ plan

Client: Cursor 3.24.18 (stable, updated Oct 14) Account: ahqonejewelry@gmail.com — Pro+ $60/mo Usage: Cursor Models 28% used, Other Models 0% used (quota NOT exhausted) Issue: All requests fail with resource_exhausted (capacity error), including a brand-new empty conversation with no attachments and a single short message. Request IDs: fd5bfa87-2043-474c-a5ef-7edc63d2a426 (modelId: grok-4.7) 437498d0-6394-49d9-9e70-fa2275966817 (modelId: claude-opus-4-8) Errors observed: grok-4.7: ConnectError…

Connection error with Claude Opus 5.5

Where does the bug appear (feature/product)? Cursor IDE Describe the Bug The IDE cannot connect to Claude Opus 5.5. I tried waiting and trying again, but it won’t connect Steps to Reproduce Interacting with AI Expected Behavior Let Claude answer Operating System Windows 10/11 Version Information Version: 3.24.18 (user setup) VS Code Extension API: 1.128.0 Commit: be4c9e5350d77747b42cd467cbd3370a753d4640 Date: 2026-10-13T18:12:58.182Z Layout: IDE Build Type: Stable Release Track: Default…

Request blocked by Anthropic

Where does the bug appear (feature/product)? Somewhere else… Describe the Bug Request blocked by Anthropic. We are unable to complete this request because it was blocked under Anthropic’s Usage Policy. Troubleshooting Already Attempted Started new conversations. Explicitly selected Claude Opus 5.5. Attempted to disable multi-agent functionality. Removed previously configured agents. Simplified prompts to request single-agent execution. Steps to Reproduce launch Claude Opus 5.5 Operating System…

Identically powered Antropic provider api question claude.ai instalation cursor instalation side by side. (?!?!)

Where does the bug appear (feature/product)? Somewhere else… Describe the Bug Title: Confused about Claude Pro vs Cursor subscription vs Anthropic API — are they sharing resources? Hey everyone, hoping someone can help me untangle this because I’m going in circles. I have a Claude Pro subscription on claude.ai (signed in via Google, account name “TEN”). I also have a Cursor Pro subscription (also signed in with the same Google/Gmail, account name shows as my Gmail display name). Both are paid…

LLM not using apply patch when generating code

Where does the bug appear (feature/product)? Cursor IDE Describe the Bug Opus 5.5 (high, 1 M) and Sonnet 5.5 (high, 1 M) deliberately used python terminal command to produce code instead of the apply_patch usual flow that generates code changes and fire the review features from cursor IDE. Steps to Reproduce Well, I don’t know, the LLM started using the usual apply patch and then just switched to the Python command-line file update flow. Expected Behavior Well, LLM, when producing code and not…

Some LLMs are not using apply_patch anymore

I was using Opus 5.5 and Sonnet 5.5 today. They started using the apply patch command for a while, then avoided it by using Python code in the terminal. This is an issue for reviewing code produced in the terminal because the IDE’s code changes feature doesn’t track it. Plus, I have no reconciliation flow to add those code changes back into the review. Am I the only one having that behavior? 3 posts - 2 participants Read full topic

Cursor SDK: long Opus 5.5 local runs end mid-turn after file reads (\"[canceled] This operation was aborted\", \"repeated resume attempts made no progress\")

Where does the bug appear (feature/product)? Cursor SDK Describe the Bug Long local runs of claude-opus-5-5 end 7 to 15 minutes in, right after file reads, without a client cancel. Other runs go silent for minutes and then carry on. We’ve seen two error texts. A. [unknown] [canceled] This operation was aborted (2026-10-04, effort=xhigh, context=1m, fast=true). The client sent no cancel: its only calls for these agents were agent.open, run.start and, well after the abort, agent.close. No other…

Anthropic down on cursor?

Cursor cant connect to anthropic API. I checked cursorAI status and anthropic status while they had an incident and stopped working now i checkede again and it says all status is operational but anthropic is still not working. All other models work fine like grok 4.7. Does any1 else have similar problems or is it just me? 4 posts - 3 participants Read full topic

Claude Opus 5.5 High not available on Ultra — only Medium shown

Where does the bug appear (feature/product)? Cursor IDE Describe the Bug I am on the Cursor Ultra plan. Claude Opus 5.5 is enabled in Settings → Models, but the Agent model picker only offers Claude Opus 5.5 Medium. Claude Sonnet 5.5 High, Claude Fable 5.1 High, Claude Opus 5 High and Claude Opus 4.8 High are all available in the same picker. Cursor’s current Claude Opus 5.5 documentation recommends the High thinking variant, and CursorBench lists Opus 5.5 High, so I expected High to be…

Repeated false-positive Anthropic policy blocks with Claude Opus 5.5

Where does the bug appear (feature/product)? Cursor IDE Describe the Bug Hello Cursor Support, I am repeatedly encountering “Request blocked by Anthropic” while using Claude Opus 5.5 in Cursor Agent. The task is ordinary software design work for a business SaaS application. It involves designing and planning a dictionary-management feature, including UI design, database/data design, and generating a static HTML mockup. It does not involve cybersecurity, penetration testing, exploitation,…

Sonnet 5.5 Subagents Route to Auto

Where does the bug appear (feature/product)? Cursor IDE Describe the Bug This is the exact same problem that occurs with Grok 4.7 subagents, as noted in the thread: Grok 4.7 selected as Subagent model but spawned as Auto instead Steps to Reproduce Exact same as Grok 4.7 Expected Behavior A subagent of the requested model should spawn, but instead it routes to Auto. Screenshots / Screen Recordings Operating System Windows 10/11 Version Information Version: 3.22.12 (user setup) VS Code Extension…