Dark Factory: OpenClaw Ships Faster Than You Can Read the Diff — Vincent Koc, OpenClaw
TL;DR
800 commits a day is already real: Vincent says OpenClaw peaked around 800 commits daily with only 10 to 15 core maintainers who all have day jobs, and he personally hit nearly 3,000 commits in a single day.
The bottleneck has shifted from typing code to managing taste: He compares the moment to the Industrial Revolution, where engineers stop being weavers in editors and start acting more like factory managers directing swarms of agents.
A 2 a.m. refactor became 2,700 commits and plugins: After a folder move broke assumptions, the team tore through nearly a million lines, touched 82 percent of the core codebase, and emerged with a plugin architecture instead of more feature bloat.
Overfit AI tests accidentally saved the rewrite: Vincent jokes that the ugly unit tests agents love to generate became the safety net, because if those highly specific tests still passed after the refactor, the team knew they were close.
The system is simpler than it looks: His setup is mostly multiple parallel coding sessions organized into swim lanes for CI, features, bugs, and triage, with Discord and GitHub signals feeding what gets attention next.
2025 was token maxing, 2026 is token efficiency: His closing claim is that success will depend less on raw model power and more on process, soft management skills, and knowing when an agent's explanations sound wrong.
The Breakdown
OpenClaw hit 800 commits a day across 10 to 15 part-time maintainers, and Vincent Koc argues that this kind of agent-driven shipping speed is about to become normal. His real point is not "just run more tokens" but build doc factories: swim lanes, harnesses, intuition for when agents are bullshitting, and the taste to say no before a codebase turns into a fire dump.
Was This Useful?
Share
Keep Reading
Make Alcreon Yours
Tune your feedFive quick questions, and the feed ranks what matters to you first.Or just get notified
The weekly Echo. Signal worth keeping in your inbox.
Every new piece, announced on X.
Read Next
See all
Playbook
Tasteful Skills
“Tasteful Skills” argues that the best agent skills are not documentation or best-practice lists.

Playbook
The Art of Tasteful Prompting
Learn how tasteful prompting helps you move beyond generic AI output by shaping context, style, and judgment from the start.

Playbook
The Codex /goal Playbook
OpenAI shipped /goal for the Codex CLI. It turns a prompt into a persisted, self-continuing contract.