13 needs clustered under "AI platform gaps" — scattered threads about the same pain, side by side, so you can compare angles.
Claude Code, Codex, Cursor, Windsurf all seem to have slightly different ways of handling project context, rules, memories and session history.
I maintain a set of REST microservices (Go and Node.js, in Docker on Linux) and our API regression suite is mostly hand-written Postman collections. Keeping them updated takes too much time, and our mocks for downstream services drift from real behaviour.
On a client project, Copilot and Codex rate-limit me for 30+ minutes at a time — even on paid plans. Looking for a free/open-source VS Code assistant with no strict caps.
We found ourselves frequently switching between different coding agents (Claude Code, Codex, Cursor) anytime we ran out of usage limits or a newer model came out. We became increasingly frustrated with having to re-explain our tasks, workflows, common mistake patterns, etc. anytime we switched agents.
Swagger presents a JSON API schema nicely. What tool does the same for a database schema stored as JSON/XML? I want to track schema changes with Git only.
I want to resolve it against a large list of known breeds (400+) and pick the closest one from the list.
Every run needs a fixed, checkable permission profile: file tools only, no hooks/MCP/plugins.
Share ideas like clearer documentation, detailed logging and better error diagnostics, expanded token limits…
We need to extract a text string from documents… 20 different formats and sometimes needs OCR. Is there an API that does this for you?
I find myself switching between ChatGPT voice mode on my phone and Codex a lot to copy and paste requirements I've brainstormed.
Converting German text to speech, even OpenAI and ElevenLabs make lots of mistakes — '74000€' and '1.' come out garbled, so users hand-rewrite the text.
High-end phones have embedded LLMs accessible via AICore, usable offline. I'm building an app on this SDK and want to crowdsource real responses across devices.
3rd parties we're working with have no way of knowing if we generated those responses ourselves… there are implications.