unlazy
Anti-laziness skill for AI agents. 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.
Unlazy is an anti-laziness agent skill that enforces completion discipline for substantial AI-driven work using the Depth Tree method and runnable verification gates. The skill splits tasks N layers deep and allocates the full time budget to every leaf, multiplying effort with depth to combat model laziness and premature completion. It provides a structured gate contract system where agents write acceptance criteria first, execute reviewed checks, and reverify work with evidence-based reporting. Grounded in 2025-2026 research on AI model underthinking and partial compliance, unlazy coordinates parallel work through scoped pipelines and optional stop hooks.
Key Features
Use Cases
- 01Preventing premature completion on multi-part refactoring tasks by enforcing verification at every decomposition leaf
- 02Coordinating parallel test execution across independent modules with file ownership leases and rolling orchestration
- 03Enforcing migration path verification in payment or checkout integrations with runnable gates and re-verification
- 04Managing complex agent work with explicit acceptance criteria written before execution begins
- 05Blocking Claude Code sessions from completing until all defined verification gates pass with evidence
- 06Auditing agent work quality through captured evidence, exit codes, and output fingerprints tied to check definitions
Related Skills
View moresuperpowers
An agentic skills framework & software development methodology that works.
skills
Skills for Real Engineers. Straight from my .agents directory.
skills
Public repository for Agent Skills
ponytail
Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.
unlazy — FAQ
What is unlazy and what does it do?+
Unlazy is an agent skill that fights AI model laziness using the Depth Tree method, which decomposes tasks into N layers and gives each leaf the full time budget of the entire task. It enforces completion discipline through runnable verification gates that require explicit acceptance criteria, command execution, and evidence-based reporting before work is considered complete.
How do I install unlazy as an agent skill?+
Use the skills CLI with 'npx skills add Leonxlnx/unlazy' for supported agents, adding '-g' for user-level or '--all' for every detected agent. For manual installation, clone the repository into ~/.claude/skills/unlazy for Claude Code or ~/.codex/skills/unlazy for Codex CLI. Invoke it with '/unlazy' (slash skills), '$unlazy' (Codex), or natural-language triggers.
Which AI clients and agents work with unlazy?+
Unlazy works with Claude Code and Codex CLI through the skills framework. The optional stop hook feature is specific to Claude Code and requires manual installation via the install-hooks.mjs script.
What are the prerequisites and dependencies?+
Unlazy requires Node.js 16 or newer for the gate checker and optional hook scripts. It uses zero third-party runtime packages. The skill itself (SKILL.md) is read by compatible agents; the gate-check.mjs script validates runnable gates and manages approval records in an owner-private directory.
Is unlazy free and open source?+
Yes, unlazy is released under the MIT License and is fully open source. The complete source code, including the skill definition, gate checker, orchestration scripts, and tests, is available on GitHub.
What is the Depth Tree method and how does it work?+
The Depth Tree method splits a task N layers deep and allocates the entire time budget of the whole task to every leaf node, so total effort multiplies with depth. This counters model laziness by ensuring that even the deepest subtasks receive full attention rather than rushed completion, backed by runnable gates that verify each leaf meets its acceptance criteria.
How do I install unlazy?+
Open the source repository on GitHub and follow its README. unlazy is a skill — MCP Agents Market links you directly to the official repo.
Is unlazy free?+
unlazy is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.