# unlazy

A completion-discipline skill for AI coding agents that requires writing an acceptance checklist before substantial work, running and reverifying its checks, and reporting only what the evidence supports.

unlazy is a skill for AI coding agents that enforces completion discipline on substantial work through runnable gates rather than a simple self-assessment of whether a task is done. Its Depth Tree method splits a task N layers deep and gives every leaf task the full time budget of the whole task, so effort multiplies with depth instead of being divided across subtasks.

A user or agent writes an acceptance ledger of gates first, each with a shell command to check it, an expected result and evidence; a checker script can inspect a ledger without executing anything, print a resolved command for review before running it, execute an approved ledger, or reverify every gate including ones already marked complete. It installs through a skills package manager into locations such as Claude Code's or Codex CLI's skills directory, and is invoked as a slash command, a dollar-command, or by natural-language trigger.

Its checker and optional hook require Node 16 or newer and use no third-party runtime packages.

Website: https://github.com/Leonxlnx/unlazy