unlazy

Overview of unlazy

What it is

Completion discipline for substantial AI-agent work, backed by runnable gates. Core: the Depth Tree method, which splits a task N layers deep and gives every leaf the full time budget of the whole task, so effort multiplies with depth. Grounded in 2025-2026 research on model laziness, underthinking and premature completion.

Key features

• Depth Tree method: splits a task N layers deep, each leaf receives the full time budget of the whole task.
• Gate contract: write acceptance ledger first, execute reviewed checks, reverify returned work, report only what evidence supports.
• Gate checker script (gate-check.mjs) with modes: --status (non-executing), normal (prints resolved command without executing until approved), --approve, --reverify.
• Security boundary: approval records stored in ~/.unlazy/approved, specific to absolute ledger and gate; editing any bound input requires re-approval.
• Orchestration and parallel work: scoped pipelines under .unlazy// with state management (WAITING, READY, IN-FLIGHT, VERIFIED, ABANDONED).
• Shell and PATH: configurable via --shell or UNLAZY_SHELL, inherits launch environment.

Use cases

Substantial AI-agent work such as refactoring code modules and verifying migration paths. Coordinating parallel work with fresh contexts using scoped pipelines.

Getting started

Install: text npx skills add Leonxlnx/unlazy Quick start: text /unlazy tree 5 refactor the payment module and verify every migration path

License and pricing

MIT license, open source.

Based on the project's public README and repository metadata.