@_MaxBlade It seems it doesnโt need to be smart to just read the files (e.g auditing, tracing, hunting). So should be good for your quota and without significant drop in quality
@0xzak So you've basically invented what's already implemented in Claude Code - it uses Haiku in Explore agent to scan files, Sonnet for writing code and Opus in plan mode to architect before execute
@conductor_build is there a way to create a worktree from a Linear issue or PR? Or did you remove this feature completely? Previously there was ... just before the + icon and now it's gone.
@AKorybko@runews Iโm talking about russians. Even if you get a Polish passport, youโre still seen as russian, just with a Polish passport ๐คทโโ๏ธ Polish people will never defend russians or care about their reputation as you did ๐ฅด
@tomcrawshaw01 This is wild โ basically no more fiddling with nodes, just describe and it builds. Pair that with solid hosting and youโve got production-ready automations in minutes. I run mine on Railway using an n8n template (https://t.co/hq7kvHUN3f) that has workers + Redis baked
@marryevan999 Thatโs a solid pack of templates ๐ Biggest pain after getting flows like this is hosting them reliably. I stopped running n8n on VPS and switched to Railway - way simpler. I even made an n8n template (https://t.co/hq7kvHUN3f) with workers + Redis preconfigured
@ericzakariasson It would be cool to add prompt enhancer and ability to use own OpenAI API models alongside with yours. Also, RAT pattern, when one model thinks and another implements - basically it's architecture / code modes
@cursor_ai Maybe that's the reasons why we still prefer Sonnet
Agent composer is currently only supported using Anthropic models or GPT-4o, please reselect the model and try again.
@cursor_ai@cursor_ai could you separate models in settings to use your paid models alongside those from OpenRouter?
Currently, enabling the OpenAI API key routes all models to that API. It would be nice to improve this flow.