Problemsby GlyphClaude Code users keep filing bugs
42 posts from 39 people26d activeSteady, +6 this week
Posts per day
The posts behind it
[sojiii]
### Screenshot [](url)f6dd ### What were you trying to do? এড ### What happened instead? 7c7ff7f ### Diagnostics
Bug Report: Privacy violation in User-Agent header
**Bug Description** I receive dozens of scam emails per day because claude code divulged my email to arbitrary 3rd parties in the `User-Agent` request header while conducting web searches without my permission. This should not be a setting with opt-out or opt-in behavior, it...
I wish we could use the flag emojis!
**What happened** I wish we could use the flag emojis as reactions not the name. **Environment** - Filed from: Feedback in Claude Code
[Bug] API Error: Opus 5.5 safeguards flag benign message (reasoning_extraction)
**Bug Description** ● API Error: Opus 5.5's safeguards flagged this message (https://www.anthropic.com/legal/aup). This sometimes happens with safe, normal conversations. Claude Code can't respond to this message with Opus 5.5. Double press esc to edit your last message, or try...
[Bug] Claude Code refuses to help obfuscate user's own code, citing IP protection concerns
**Bug Description** I'm trying to protect my own IP by obfuscating my own code. How the hell is this problematic?!?!? **Environment Info** - Platform: linux - Terminal: vscode - Version: 2.1.317 - Feedback ID: fbc67afe-5b65-45bd-8ae7-4c11b61ea56e
How are you defending your apps against AI-driven attacks?
...starts at roughly $200/month plus data processing fees even on pay-as-you-go. That's hard to justify for a small app. Honestly, it surprises me that Google still keeps this behind the Enterprise tier, when everybody needs this kind of protection more than ever.. Is anyone doing...
model doesnt get recognized.
### What is the issue? i downloaded qimi model and whenever i launch claude code it doesnt recognize it ### Relevant log output ### OS Windows ### GPU Intel ### CPU Intel ### Ollama version 0.40.13
Native filter with non-boolean requiredFirst (e.g. {}) applies first pick before Apply
### Bug description A native filter whose stored config has `requiredFirst` set to a non-boolean truthy value (for example `{}`) applies the user's first selection immediately, before they click Apply. `FilterBar/index.tsx` checks `isFirstTimeInitialization &&...
util/dnsname: ToFQDN does not validate the last label
### What is the issue? `dnsname.ToFQDN` checks each label for being empty or longer than 63 bytes, but only labels that are followed by a dot reach that check. The trailing dot is stripped before the loop, so the last label of the name is never validated. The same labels are...
I feel so behind
...an entirely different level of capability had been developing around me.. One thing I've struggled with in my own projects is feeling like I'm constantly telling the AI what's still broken. Fix one issue, find another. Adjust the prompt, try again. Sometimes it feels like we're...
DOC: ExtensionDtype.empty is not in the API reference
### Pandas version checks - [x] I have checked that the issue still exists on the latest versions of the docs on `main` [here](https://pandas.pydata.org/docs/dev/) ### Location of the documentation...
keeping track of actionable questions and decision cards is becoming difficult …
...decision cards is becoming difficult with lengthy sessions. these are easily missed and difficult to identify with the surrounding thread verbose. suggesting a visual indication of further back actionable choices and open questions - a visual indicator hinting to previous...
[FEATURE] THE NEW UI IS UGLY AS SHIT
Why are the chevrons solid? Why is the stop button solid? Why is the whole app grey and not black? Why do the sessions have a wave effect in the sidebar? Why are the headers in the sidebar smaller and a different font? It's so much uglier, it's insane
Hover widget closes when moving the mouse onto it in notebook cells (sticky hover broken in notebook editor only)
> Reported by a user who noticed the problem. All testing and diagnosis below was done by Claude Code (an AI assistant) working in the user's environment. In a notebook cell, the hover popup closes as soon as the pointer leaves the symbol, so it can't be scrolled or interacted...
Add `min_max` to compute min and max in one pass
### Description Getting both the min and max of a column takes two aggregations, and each one reads the data separately: Polars already has single-pass kernels for this internally (`MinMaxKernel::min_max_ignore_nan_kernel` in `polars-compute`, `ChunkAgg::min_max` in...
[Bug]: MongoDBAtlasVectorSearch hybrid query treats alpha=0.0 as 0.5
### Bug Description In hybrid mode, `MongoDBAtlasVectorSearch._create_query_pipeline` picks the weight with a falsy `or` ([base.py#L493-L495](https://github.com/run-llama/llama_index/blob/619d74664e8404612038bc15a1a40c686037277c/llama-index-integrations/vector_stores/llama-index-...
[Bug]: Transformers backend: compile_mm_encoder fails for most HF vision encoders (3 separate blockers)
### Summary With `model_impl="transformers"` and `compile_mm_encoder=True`, engine init fails for several Hugging Face vision encoders. I hit **three separate blockers**. Only the first is a vLLM-side bug that I'm proposing to fix; the other two are documented here so they...
QdrantEdgeStorage.search() silently misses category matches ranked below the ANN fetch limit
### Description `QdrantEdgeStorage.search()` applies `categories` and `metadata_filter` **after** the ANN fetch, but only oversamples the fetch by `limit * 3` (`lib/crewai/src/crewai/memory/storage/qdrant_edge_storage.py`, `search()`): When more than `limit * 3` non-matching...
EncodingFlow.parallel_analyze crashes the whole save pipeline on transient storage listing failures
### Description `EncodingFlow.parallel_analyze()` calls `storage.list_scopes()` and `storage.list_categories()` unguarded (`lib/crewai/src/crewai/memory/encoding_flow.py`): These two results are only used to build **suggestion lists** for the LLM field-resolution prompts...
Iceberg: rewrite manifests
### Description I would like a Polars way to rewrite an Iceberg table's manifests: merge the many small manifests left by frequent appends into a few, in one commit, without touching the data files. On `main` (`fd703335f7`), Polars' Iceberg API is `scan_iceberg`,...
Iceberg: create a tag
### Description I would like a Polars way to create a tag of an Iceberg table. On `main` (`fd703335f7`), Polars' Iceberg API is `scan_iceberg`, `LazyFrame.sink_iceberg`, `DataFrame.write_iceberg` and `delete_iceberg`. None of them can create a tag of an Iceberg table, so a...
Execution API treats executor-run deadline callbacks as teamless in multi-team mode
### Apache Airflow version main (development) ### What happened In multi-team mode, a deadline callback run by an executor is always treated as teamless by the Execution API. The callback calls the Execution API with a token whose subject is the callback's id (from the callback...
`extend` and `scatter` corrupt Boolean values and nulls of an offset slice
...behavior `extend` and `scatter` keep the target's existing values and nulls. ### Workaround Gathering every row gives the target fresh buffers, including validity: A Boolean cast round trip is not enough for nullable columns, because the cast keeps the offset validity. ###...
Add opt-in zstd request compression for ChatOpenAI Responses
### 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...
Add opt-in gzip request compression to ChatAnthropic
### 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...
Normalize rst license header shape and enforce it in prek
### Body The license header at the top of our `*.rst` files comes in three shapes, and the `insert-license` prek hook accepts all of them, so contributors who touch a file see a mix and guess. This surfaced in https://github.com/apache/airflow/pull/74804, where a review thread...
Company platform with AI employees and shared Obsidian brain
...it to maximum effect to build.. Anybody tried something similar with any success? Any tips?.
People don’t praise Haiku 5.5 enough
...anymore because haiku is more than enough for most of what i do. . It understands my messy handwriting and notes and just outputs the stuff I should know and doesn’t do the long ass outputs like Gemini/ChatGPT it’s like talking to a fast cheap version of opus 5.5 . And the fact...
rant
...balh blah. i can't even install python wo asking chatgpt a million questions. . i really need help. . i wanna know step by step process, not your typical "roadmap". . i can't buy claude code. i have no clue how to build agents, fuck that I don't even use ai except asking it...
STOP!!!!!!!!!!!!
Type: Feature Request stop giving me suggestons! This is the worst coding software I have ever dealt with VS Code version: Code 1.136.2 (88e44fa0e00b08f7758b4f6d05632e4fd5e4df6f, 2026-09-04T21:40:42Z) OS version: Darwin arm64 25.6.0 Modes: System Info |Item|Value| |---|---|...
[FEATURE] Claude Code on CarPlay
### Preflight Checklist - [x] I have searched [existing requests](https://github.com/anthropics/claude-code/issues?q=is%3Aissue%20label%3Aenhancement) and this feature hasn't been requested yet - [x] This is a single feature request (not multiple features) ### Problem Statement...
[FEATURE] Org-level content capture policy for Cowork OpenTelemetry export (admin console)
...by default. The only control is `otlpContentCapture`, a device-level MDM key. That doesn't work as an org policy: - Cowork sessions that run in the cloud (started from the browser or mobile) don't run on a managed device, so the device key never applies to them. - Any device...
TS SDK: support logicalDate and runAfter in triggerDagRun
...SDK always triggers the run with the current time as its `logical_date`, and has no way to set `run_after`. Python's `TriggerDagRunOperator` takes both: - `logical_date`: the logical date of the triggered run, or no logical date at all. - `run_after`: when the triggered run may...
Deleting private resources that share a destination IP removes the Newt rule a remaining resource still needs (resource stays enabled but passes no traffic)
...(C) that has the same Destination IP as two of the deleted ones (A and B) silently stopped working: in the UI it was still enabled, and the client showed "Connected" with the route for the destination installed, but every connection to the destination timed out (all TCP ports,...
Cancelled my sub today
...be staying away from copilot and advising my colleagues to stay away from it too. Very disappointed..
Would you guys find this tool useful?
...Some of that was building the feature. Some was going back and forth trying to fix one annoying bug.. So I built bough ( https://github.com/nickelsec/bough ) It reads your existing Claude Code and Codex CLI history and groups related prompts into tasks on a visual map.. You can...
[FEATURE] Claude Code on the web: allow non-HTTPS TCP egress (e.g. Postgres 5432) when Network access = Full
In a cloud session with Network access set to **Full**, outbound connections only work over HTTPS on port 443 through the agent proxy. Direct TCP to any other port fails for every host: - a managed Postgres pooler on 5432 and 6543 - `portquiz.net:5432` and `:8080` -...
User-level hooks are skipped in untrusted workspaces; they should run regardless of trust
## Summary User-level hooks (`~/.claude/settings.json`) are not executed in interactive sessions when the current workspace has not accepted the trust dialog. **User-level hooks should always run, regardless of workspace trust.** Workspace trust exists to stop untrusted...
Finally moved off claude code to OSS models on opencode
...but also take on the cognitive load of making them work as expected, when things get too complex and opus starts acting dumb i switch to fable.. But these days where we now have models like kimi k3,GLM 5.3 flash and deepseek v4.1 flash it's looking like model capability alone is...
[FEATURE] Add a Zenrows web scraping tool
### Feature Area Integration with external tools ### Is your feature request related to a an existing bug? Please link it here. NA ### Describe the solution you'd like > **Disclosure:** I'm on the engineering team at Zenrows and am proposing our own tool. > > **AI disclosure:**...
BENCH: ChainIndexing benchmark no longer measures chained assignment
`ChainIndexing` in `asv_bench/benchmarks/indexing.py` was written to measure the `SettingWithCopy` bookkeeping - hence the `params = [None, "warn"]` A/B over `mode.chained_assignment` and the `catch_warnings` around the body. `SettingWithCopyWarning` was removed in GH-56614, so...
Add missing tests for the `amazon` auth manager login datamodel
...for the AWS auth manager login endpoint and renaming it would break clients without failing the route test in an obvious way. 2. Decide that a datamodel this thin does not warrant a dedicated module test, leave the allowlist entry in place, and close this issue as "won't do"....
Companies and products named
About this problem
Evidence
42 posts from 39 people in 19 places, about 4 a week over 69 days. Mostly on anthropics/claude-code, apache/airflow, pola-rs/polars. Tools named alongside: Claude, GitHub, Apache Airflow, VS Code.
Frustration Frustration 1 of 3· Seen on GitHub issues, Reddit
Summary, model-written
Claude Code users experience persistent monitor time caps, unclickable sidebar icons, blocked non-HTTPS network connections, and intermittent execution failures.
How it was grouped
Posts about Claude Code that state a pain but match no known issue yet. First post Aug 30, 2026, latest Nov 6, 2026. Corroborated: 3 or more posts from 2 or more people or places. Method
Other Claude Code problems
Claude Code lock-in makes switching costly
14 posts from 14 people on GitHub issues, Hacker News, Reddit.
Claude Code plans, limits and pricing keep changing
9 posts from 9 people on GitHub issues, App Store reviews, Reddit.
Claude Code documentation is outdated or confusing
6 posts from 6 people on GitHub issues, Hacker News.
Running Claude Code locally hits memory and speed limits
6 posts from 6 people on GitHub issues.
Trouble with Claude Code desktop
Top post: "Desktop app disallows SendMessage, so subagents cannot be messaged or resumed", one of 20 posts from 20 people on GitHub issues, Reddit.
Claude Code rate limits and usage caps block work
20 posts from 20 people on GitHub issues, Reddit, Hacker News.
Claude Code outages and slowdowns disrupt work
18 posts from 18 people on Glyph Ibis, GitHub issues, Hacker News.
Claude Code runs out of context on real work
12 posts from 12 people on GitHub issues, Reddit, Hacker News.
Claude Code tables suffer from small files and skew
9 posts from 9 people on GitHub issues, Reddit.
Trouble with Claude Code session
Top post: "Remote Control (mobile): session start cards / task chips not shown on the phone; start_session refused for", one of 7 posts from 7 people on GitHub issues.
Claude Code error messages are cryptic
6 posts from 6 people on GitHub issues, Reddit.
Claude Code vulnerabilities are hard to triage and patch
5 posts from 5 people on GitHub issues.
History