Products · Product
Claude
Raised as a usability problem by 310 people on Hacker News over 11 years (2016–2026), in 243 threads.
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
Every issue, all time, most people first
Responses are cluttered with excessive verbose text, hard to readUnreviewed
20162026- Responses are overlong and jargon-laden, so the user must read everything to follow suggestions2026-09-18
- Responses are too verbose and lack typographic hierarchy, so they are hard to scan2026-08-21
- Model's responses are so verbose and padded that it is borderline unusable for conversation2026-08-21
Chat lags badly because it re-renders entire conversation each updateUnreviewed
20162026- Chat becomes almost unusable in Firefox after recent changes2026-03-31
- Chat UI lags badly because it re-renders the entire conversation on each update2026-03-29
- Loses chats when switching tabs, and is slow2026-01-07
New tasks hang for seconds, making rapid switching intolerableUnreviewed
20162026- Fails to respond in long chats with no error message shown2026-04-10
- Switching conversations takes seconds, and no progress indicator shows when work is happening2026-03-03
- Long conversations break rendering and the app feels janky and slow to start2026-03-03
Too many requests to Opus 5 cause a poor experienceUnreviewed
20162026- Sends too many requests to Opus 5, creating a poor experience2026-09-01
- Silently falls back to Opus, so users don't know which model they're using2026-09-01
- Gets switched back to Opus after about two minutes of a task2026-07-08
Forgets repeated instructions minutes later, so users must re-remind itUnreviewed
20162026- Keeps answering with what things are not, even after repeated reminders, forcing repeated corrections2026-09-13
- Ignores standing writing rules and memories mid-writing, so the user must repeat instructions constantly2026-08-24
- Skill instructions get ignored within a few messages, forcing repeated correction2026-08-14
No cancel option when subscription is paused, so user can't cancelUnreviewed
20162026- Can't figure out how to switch subscription type without cancelling and resubscribing2025-10-05
- Cancel option missing from account when subscription is paused, so user can't cancel2025-09-30
- Can't remove payment method or get support after subscribing2025-09-18
Email magic link login is frustrating every timeUnreviewed
20162026- Email magic link login is a frustrating experience every time it is used2026-09-18
- Account email cannot be changed, so users are stuck with the email chosen at signup2025-12-20
- Gets logged out often and must wait for email code every login2024-10-01
Map results buried in chat history and not interactiveUnreviewed
20162026- UI randomly hides previous messages from visible chat history, though the model still has them2026-07-24
- Built-in map results are buried in chat history and not interactive2026-06-21
- Historical chats take 5-10 seconds to load, and history is only searchable by irrelevant auto-generated titles2025-07-18
Signed up with Apple on phone but desktop only offers Google sign-inUnreviewed
20162026- Doesn't work when signed into multiple Google accounts, so documents can't be accessed2026-10-23
- Sign-in takes three attempts and long waits when switching Google accounts2026-08-28
- Website lacks Sign in with Apple, though Google sign-in is offered2026-08-18
Agent makes destructive mistakes unpredictably, wasting users' timeUnreviewed
20162026- Agent gets stuck on arbitrary restrictions and makes user do tasks manually2026-09-28
- Agent tool makes destructive mistakes unpredictably, so users get angry and wasted time2026-05-26
- AI agent is too chatty, annoying to interact with2026-04-01
Turn-based chat is tedious and unproductive for real workUnreviewed
20162026- Dumps distracting tool-call cards in chat without explaining why, making it hard to follow2026-10-22
- Chat interaction is tedious and turn-based, making it unproductive for real work2026-09-16
- Users cannot see or change the reasoning effort level their chats get2026-04-10
Can't make Claude restart or clear resources automatically when context runs lowUnreviewed
20162026- Claude degrades past about 300k context and compaction doesn't fix it2026-07-18
- Claude often stalls, forcing constant context switching while waiting on it2026-03-31
- Can't make Claude restart or flush resources automatically when context runs low2026-02-11
Unreviewed issues were grouped by machine: Claude Haiku labelled each comment as a usability complaint, named what it's about and summed up the problem in a line, and a small model grouped similar problems. Nobody has checked them by hand. Expect mistakes.
The newest 50 complaints
- Automated tests need a fragile manual token extraction process from the browser2026-11-01
- Spams the UI with explanatory text for every control2026-10-30
- Stops a code security scan partway without saying what it found or why2026-10-29
- Keeps pushing user toward definitive decisions instead of open exploration2026-10-28
- Feedback button is dark-patterned, so the user avoids it2026-10-27
- Doesn't work when signed into multiple Google accounts, so documents can't be accessed2026-10-23
- Generated UIs are broken or ignore guidelines, and revisions alter the entire design2026-10-22
- Dumps distracting tool-call cards in chat without explaining why, making it hard to follow2026-10-22
- Makes user retype information already known to the signed-in chat session2026-10-21
- Constantly refused on simple tasks, with lectures instead of help2026-10-13
- Answers are too long and padded; system prompts get forgotten and concise mode is insufficient2026-10-07
- Responses feel overly verbose and slow, and large shared config docs slow everything down2026-10-06
- Simulation takes forever to load and runs poorly2026-10-04
- User wasted minutes hunting for the update option, which on Linux must be done via apt2026-09-28
- Assistant won't edit a settings file, forcing the user to do it themselves2026-09-28
- Agent gets stuck on arbitrary restrictions and makes user do tasks manually2026-09-28
- Side-quests off on tangents when something goes wrong instead of asking for guidance2026-09-25
- Feature missing from the app but findable on the web interface2026-09-23
- Interface scrolling is so bad it gets in the way of reading content2026-09-22
- Table of contents only appears after scrolling, and users must keep scrolling past the hero to reach text2026-09-22
- Ignores instructions, so user wastes time asking it to shorten or simplify responses2026-09-22
- Page is over-animated and endlessly distracting, making the post hard to read2026-09-21
- Responses are overlong and jargon-laden, so the user must read everything to follow suggestions2026-09-18
- Generated sites are dark-only with hard-to-read fonts, colors and sizes2026-09-18
- Email magic link login is a frustrating experience every time it is used2026-09-18
- Tracking work in progress across long chats means endless scrolling to pick up a thread2026-09-17
- Prompts now get routed automatically, forcing admins to rewrite their usage guides2026-09-16
- Normal users can't tell whether to pick chat or cowork2026-09-16
- Chat interaction is tedious and turn-based, making it unproductive for real work2026-09-16
- Confusing mix of separate surfaces; non-technical users can't tell which works for a task2026-09-16
- Toggling between chat and Cowork modes is a terrible experience; the model should pick power level itself2026-09-16
- Sidebar has unneeded items always taking prime space; can't trim it to what the user uses2026-09-16
- No slider controls for tuning the model's writing voice; user must fight the prose2026-09-14
- Keeps answering with what things are not, even after repeated reminders, forcing repeated corrections2026-09-13
- Decision prompts take over the whole screen, disrupting the user's workflow2026-09-13
- Opt-out setting keeps getting switched back on, with no way to make it stick2026-09-10
- Feedback prompt's buttons are designed so users click it and unknowingly consent to data training2026-09-08
- Training opt-in toggle turned on silently via re-consent modal, unnoticed2026-09-08
- The interface looks like generic AI-generated output, which the commenter finds off-putting2026-09-04
- Checkout through the chat interface is awkward for users2026-09-03
- Answers only one of two yes/no questions and adds unrequested answers, hard to parse2026-09-02
- Agent reports only summarized back, with unexplained jargon and missing context, forcing repeated prompts2026-09-02
- Final summary is inscrutable; user must work hard to find the actual problem2026-09-01
- Slow, and it grabs application focus while the user is working elsewhere2026-09-01
- Sends too many requests to Opus 5, creating a poor experience2026-09-01
- Silently falls back to Opus, so users don't know which model they're using2026-09-01
- Rewritten UI labels and error messages are pompous, odd, and harder to understand2026-09-01
- Sudden redesign pushed the user into a mode they never use2026-08-31
- Sign-in takes three attempts and long waits when switching Google accounts2026-08-28
- Responses take 5-7 minutes per turn, too slow to use interactively2026-08-24