51 real-world AI agent setups analyzed: narrow tasks, drafts not sends, file-based handoffs
abe17124 · reddit · 2026-10-18
The author curated 51 agent setups that people actually keep running, traced to real users. Highlights: Boris Tane has Claude Code produce research.md then plan.md with up to 6 human-annotated rounds before coding; Wyndo/Dheeraj Sharma run Claude Code and Codex sharing a changes.log so the second agent reviews the first; Noah Hailegiorgis's n8n inbox triage cut email time from 3.5h to 30min a day. Common threads: narrow jobs, drafts instead of sends and PRs instead of merges, handoffs in files, separate reviewer agents, quiet-until-threshold monitors.
More from coding & agent
- Google's Deep Review Tore a 35-Page Benchmark Spec Apart Like a Hostile NeurIPS Chair — ChrisUniverse · 2026-10-18
- Dev open-sources an X network map tool with a Grok agent prompt to collect data — jacob_posel · 2026-10-18
- Serverless, explained with a pizza shop and ghost kitchen analogy — thisiskp_ · 2026-10-18
- Pydantic AI CodeMode speculatively runs tools mid-generation, cutting 3 lookups from 8.7s to 4.7s — samuelcolvin · 2026-10-18
- 18 AI engineering concepts, from RAG variants to LoRA/QLoRA, mapped to open-source repos — MaryamMiradi · 2026-10-18
- Bilibili Creators Used Claude to Make an Undertale Music Video, Surprising Reddit — Luciaka · 2026-10-18