Vincent Koc recounts a large OpenClaw refactor and describes the role tests played as maintainers moved toward a plugin architecture. He presents the project’s reported commit volume as an example of its rapid development.
He organizes parallel coding sessions into separate work streams for tests, features and urgent issues, and discusses the practical cost of maintaining many worktrees and running heavy tests.
Koc also describes reusable skillsAn AI agent skill is a reusable package of instructions, resources, and tool guidance for performing a bounded kind of work., pull-request triage and evaluationEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria., arguing that supervising agent output requires judgementHuman judgment in AI is the accountable interpretation and decision-making people contribute when setting goals, evaluating evidence, and managing consequences. and communication alongside tools.
Watch on YouTube
