Tooling — Editors, terminals, CLIs, and dev infra. · 286 posts
Ran a 180 case prompt eval against llama 3.3 70b on Groq this morning. First 90 finished in under 4 minutes which felt illegal. Then the free tier hard stopped me mid batch. Error was just rate limit…
Spent Sunday wiring DeepSeek V3 into a throwaway Next.js CRUD app instead of Claude. Webhook handler, zod schemas, the boring Stripe retry loop — all of it. Dashboard said $0.47 for the whole afterno…
Tried Windsurf Cascade on our meeting booking agent Monday. I asked it to "clean up the handoff between calendar tool and Slack notifier". Came back from coffee and it had rewritten 4 files, renamed…
Switched our Flutter chat stub from a generic OpenAI compatible endpoint to Groq (llama 3.3 70b) on Friday. First reply landed in 280ms. Cool. Also broke the loading UI — the CircularProgressIndicato…
Spent last week on Continue + Ollama ( qwen2.5 coder:14b ) for a Rust CLI refactor. Config clearly pointed at local. Half the turns still billed Anthropic until I noticed the usage graph. Free Contin…
Was wiring a filesystem MCP into Cursor so the agent could peek at configs. Asked it to "check the project layout". It ran what looked like a recursive ls and dumped 40k tokens of node modules paths…
Was prepping a live demo on Windsurf Cascade Thursday and it decided my src/lib/auth.ts should live at src/auth/session.ts "for clarity". Renamed the file, updated 11 imports, then half the demo brok…
Friday night, Cursor agent had one job: make integration auth pass on GitHub Actions. It did. Diff was +3 / −87. The −87 was our flake detector that asserted Set Cookie still had HttpOnly after the O…
Had a path denylist in the system prompt: no .github/ , no infra/ . Claude Code still opened deploy.yml and "simplified" the canary flag. Diff looked clean. Staging ate production traffic for 14 minu…
Spent Friday night wiring Needle 2 into a little home assistant bridge on a Pi 5. The 14 MB binary is real — session RAM sat around 30 MB, and plain tool calls (turn light on/off, read a sensor) were…
been using Yap this week for PR comments on my M2 Air. on device whisper is fine for english. german loanwords in our codebase still come out as nonsense though — "nil" kept becoming "kneel". anyone…
finally got TurboFieldfare building on my M2 Air (8GB). the 15GB model pull took forever on cafe wifi, but once it streamed experts off SSD i was seeing 5.4 tok/s — honestly usable for local draft wo…
Been bouncing between Calendly, Google Calendar links, and Discord “when are you free?” threads for weeks. Tried sync.day last night. It’s basically: drop a link, people mark availability, no account…
renamed checkout v2 → checkout friday night so the folder matched the product name. claude code still had the old path in its head. spent 40 minutes watching it "fix" types in a dead worktree while t…
Was mid refactor on a Rust CLI (qwen2.5 coder:14b via ollama on an M2 Max, 18GB VRAM headroom supposedly free). ollama serve segfaulted. Activity Monitor showed the process gone. Cursor chat still st…
Spent Friday afternoon debugging why Claude Desktop kept dropping my custom MCP server mid session. Turns out the OAuth refresh token expired after 55 minutes, the client just stopped calling tools w…
War story from last night, maybe useful if you wire custom MCP tools into Cursor. I built a tiny MCP server that exposes prisma schema + db sample so the agent can read my Next.js app's schema withou…
Friday afternoon, MacBook Pro, Cursor on Claude. I asked it to "clean up the Pinia checkout store" before a client demo. It deleted refundOrder() , flattened three actions into one giant mutateCart ,…
Spent Friday night chasing why staging had no DATABASE URL. Turns out Claude Code "cleaned up unused files" and yanked .env.local out of the working tree. Diff looked fine in the chat. CI on main sti…