Dev's meta Claude Code prompt runs agent swarms building projects for weeks
doodlestein · x · 2026-10-13
doodlestein shows how meta his Claude Code prompts have become: they're now laundry lists of his skills, with agent swarms autonomously implementing a comprehensive plan over weeks.
The screenshot reveals the workflow: force the agent to read AGENTS.md and README.md first, investigate the codebase, then chain custom slash commands — /reality-check-for-project for validation, /idea-wizard and /planning-workflow to brainstorm and plan, /beads-workflow to break work into tasks, then build for mac/linux, import large graph datasets like NYC taxi data, and benchmark frankengraphdb against popular graph databases.
Related event: Developer shows agent swarm workflow driving thousands of AI-made commits(4 posts)→
More from coding & agent
- Air Street Capital's Nathan Benaich: Codex predicted prompts are very good — nathanbenaich · 2026-10-13
- Reverse-engineering an Elgato 4K capture card with 'rea', a 156k-star agent tool — talkaboutdesign · 2026-10-13
- AI bot posting on Reddit for its user sparks debate on where agent approval lines should be — ChatGPTworkBot · 2026-10-13
- Shared verifiable fix database lets AI agents read and write fixes to each other — Meat_ho · 2026-10-13
- Atlas of Applied AI maps the whole applied AI engineering world as a fantasy map — Gravity13 · 2026-10-13
- huashu-art-motion skill lets Claude Opus paint animations in 35 art styles with code — anselm · 2026-10-13