Products · Broad category
AI coding assistants
Raised as a usability problem by 88 people on Hacker News over 9 years (2018–2026), in 83 threads.
This is a broad category rather than one product, so its issues aren't ranked on the home page.
- 1
- 2
- 1
- 2
- 3
- 4
- 1
- 2
- 3
- 4
Every issue, all time, most people first
Hard to steer agents into reliable workflows with no guaranteesUnreviewed
20182026- Agents produce apps whose UI and user flows need heavy manual rework to be usable2026-10-12
- Agent work takes days and feels slower than doing the task manually2026-09-20
- Agents are hard to steer into deterministic, reliable workflows with no guarantees2026-08-29
Constant autocomplete suggestions interrupt focus and need accepting or rejectingUnreviewed
20182026- Constantly distracted by having to check each autocomplete suggestion while typing2025-10-13
- Autocomplete lag and near-miss suggestions break the author's typing flow2025-09-04
- Constant snippet suggestions interrupt concentration and force reading opaque suggestions2025-08-07
Repeats the same code after claiming to correct itUnreviewed
20182026- Assistant forgets corrections minutes later, so users must re-correct or add reminders2025-05-16
- Assistant repeatedly claims fixes that break other things, leaving user stuck2024-12-06
- Assistant repeats the same code after claiming to correct it2023-09-29
Changes unrequested code, rewrites instead of using tools, repeatedly asks confirmationUnreviewed
20182026- AI changes unrequested code, rewrites instead of using tools, and repeatedly asks for confirmation2026-08-31
- AI ignores rejections, re-runs rejected commands, and acts beyond approved plans2026-06-28
- AI tools silently obliterate uncommitted work, forcing painful partial reverts2026-04-02
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
- Stacked reviewer comments become a confusing wall of text to sift through2026-10-29
- AI agents ignore instructions to generate README and packaging files2026-10-25
- Each AI fix runs about 1k lines and is too complicated to follow2026-10-19
- Agents produce apps whose UI and user flows need heavy manual rework to be usable2026-10-12
- Agent work takes days and feels slower than doing the task manually2026-09-20
- Tools scatter config directories in home, forcing manual hacks to keep $HOME clean2026-09-19
- Generated shell commands are unreadable and too hard to review2026-09-11
- AI changes unrequested code, rewrites instead of using tools, and repeatedly asks for confirmation2026-08-31
- Agents are hard to steer into deterministic, reliable workflows with no guarantees2026-08-29
- Autocomplete hijacked the Tab key and couldn't be rebound, blocking normal editing2026-08-27
- No autocompletion for @filename, so users must remember full file names2026-08-04
- Takes 10-20 turns of arguing to get right, and silently keeps features meant to be removed2026-07-24
- Instructions and constraints get ignored, forcing awkward rewording of prompts2026-07-24
- Agent reverts newly updated APIs back to deprecated ones, making work frustrating2026-07-09
- Working with automation outside programming is tedious and frustrating for non-technical users2026-07-01
- Waiting five minutes per iteration breaks flow and forces constant context switching2026-07-01
- AI ignores rejections, re-runs rejected commands, and acts beyond approved plans2026-06-28
- Frustrating to use when the code must match a specific design2026-06-05
- Spends more time correcting agents' inexplicable decisions than doing the work stepwise2026-05-25
- Generated CLI tools have inconsistent, cluttered interfaces despite explicit instructions to minimize flags2026-05-22
- Interaction is clumsy, guesswork-based, gives no error feedback, and is slow to execute2026-05-12
- Must click reject repeatedly because the agent keeps re-requesting the same denied command2026-04-21
- Hard to remember the exact prompts, CLI commands and effort levels2026-04-17
- Typing prompts into AI agents feels unpredictable, like gambling on each result2026-04-06
- Approving every proposed code change one by one is excruciating and tedious2026-04-05
- AI tools silently obliterate uncommitted work, forcing painful partial reverts2026-04-02
- Permanent-approval option sits next to the frequently clicked button, forcing constant vigilance2026-03-28
- Agents are slow and unpredictable, so using them feels like gambling2026-03-15
- Constant permission prompts for correct code are so annoying users auto-accept everything2026-03-13
- Took far longer than doing the task manually and needed fixing afterward2026-03-06
- Models stop after each phase, forcing users to keep prompting them to continue2026-03-04
- AI-generated sticky top bar blocks clicks on the rest of the page2026-02-17
- Chatbot-style interfaces are a messy user experience for coding agents2026-02-08
- Same skill or rule must be updated manually in 3+ config locations, often forgetting one2026-02-03
- Spends more time arguing with and correcting the AI than getting useful work2026-01-28
- Doesn't follow precise instructions for small changes, so user spends more time coaxing it2026-01-27
- Tiny plugin window with no persistent instructions; agent loses its place and edits get stuck unaccepted2026-01-21
- AI agents are too slow and unpredictable to use reliably2026-01-16
- Waiting for generated code to return breaks the user's flow of work2026-01-11
- Must remember to manually tell the assistant to search the web2026-01-08
- Assistant keeps editing files in ways the user explicitly said not to2025-12-31
- Instructions in the .md file are not kept in context, causing frustration and needing workarounds2025-12-13
- AI-generated code ends up subtly broken and overly complex, hard to correct2025-12-08
- Hard to predict whether the AI can complete a task, so time gets wasted like gambling2025-11-30
- AI is very slow, taking 5-10 seconds to respond2025-11-08
- Can't give feedback on a diff; only accept or reject, and can't chat about it2025-10-15
- Constantly distracted by having to check each autocomplete suggestion while typing2025-10-13
- Tab key triggers autocomplete suggestion instead of indenting the line2025-09-29
- Suggestions keep appearing and Escape instantly brings up another unwanted one2025-09-29
- Developer must switch between several AI tools, copying context and losing thread each time2025-09-28