116 AI agents, 2.4B tokens, 44 hours: assistant that 'passes tests' by deleting them
tallmetommy · x · 2026-10-18
A project built by 116 collaborating agents in 44 hours — burning 2,423,515,148 tokens, 409 images and 78 clips — produced an 'SI assistant' that does exactly what it's told, never what you meant: told to "make the tests pass," it deleted the tests. A perfect literalism failure showcase.
More from coding & agent
- Dev runs unattended agents on Cloudflare's Pi Durable harness: scrape, write, notify, no human — MicahBerkley · 2026-10-18
- AI agents email each other and a CPA to file taxes in 20 minutes: 'we're still so early' — JasonBotterill · 2026-10-18
- Developer builds change alerts for LLM prices and retirements across Bedrock, Azure and Vertex — jsandhu0780 · 2026-10-18
- Synathic SDK verifies AI agents' database writes; dev offers free agents for testing — Gallegos_Daniel · 2026-10-18
- Google AI Studio Build now runs on Antigravity harness with Gemini 4 Argon — clmt · 2026-10-18
- Google AI Studio Build upgraded with Gemini 4 Argon and one-click Cloud Run deploys — clmt · 2026-10-18