How OpenClaw cut CI runner-hours by 86.5% and compute cost by 95.5%
vincent_koc · x · 2026-11-17
Vincent Koc's deep dive on fixing OpenClaw's CI: with coding agents, the bottleneck wasn't tokens, memory or disk but tests — agents generated more tests, jobs and retries. Weekly runner-hours fell from 158,546 (September peak) to 21,395 (-86.5%), estimated compute value down 95.5% (Blacksmith-sponsored). Key moves: remove weak tests and redundant setup, balance machine workload, use adaptive test selection with full suites only when affected-test selection is uncertain. Includes a copy-paste prompt so your own agent can audit and optimize a repo's CI.
Related event: OpenClaw CI costs cut 95.5% as tests, not compute, prove the bottleneck(2 posts)→
More from coding & agent
- Opensteer: open-source local Mac agent powered by your Claude subscription — dinotimm · 2026-11-17
- Fable 5.2 spent 7 hours and 8 critic rounds building a drivable Fiat Uno game in one 603 KB HTML file — Viktorkav · 2026-11-17
- Poland's engineering teams have stopped writing code by hand entirely — PawelHuryn · 2026-11-17
- Fixing multi-user agent collaboration: one-click rooms debut in four-part release — thechrisperry · 2026-11-17
- Open-source Pascal Editor brings near-native 3D building design to the browser via WebGPU — anthara_ai · 2026-11-17
- tsrs hits 5M weekly npm downloads in 23 days; peers took years — cnakazawa · 2026-11-17