Because AI Agents need systems, not just instructions.
"You do not rise to the level of your goals. You fall to the level of your systems."
— James Clear
Atomic Habits Skills is an open-source Claude Code plugin that retrofits the world's most popular behavior-change framework directly into your AI Agent's workflow.
Instead of treating Claude as a black box where you hope for the best, these 9 slash-command skills inject deliberate, systemic discipline (like Test-Driven Development loops, CI/CD tracking, and defensive personas) straight into the agent's cognitive engine.
Type these commands in your Claude Code terminal to immediately shift the agent's behavior for the current task:
| Skill | Agent Application |
|---|---|
/habit-stack |
Chain agent operations (e.g., "after write_to_file, always lint") |
/environment-design |
Workspace engineering: Audit project structure, surface configs, enforce .gitignore |
/implementation-intent |
Execution contracts: Pre-commit to tool chains before acting (e.g., "I will grep_search before editing") |
/two-minute-rule |
Gateway actions: Beat scope-paralysis with a simple view_file_outline first |
/temptation-bundle |
Need → Want Gating: Force tests before features ("write test BEFORE new endpoint") |
/habit-track |
CI Tracking: Track coverage, lint scores, and enforce the "Never Regress Twice" rule |
/identity-shift |
Persona adoption: "Security Paranoiac", "Scalability Hawk", "Clean Coder" |
/reflect-review |
Metacognitive audits: Sprint reviews, alignment checks, and workflow scorecards |
/goldilocks |
Complexity calibration: Detect agent boredom (sloppy CRUD) vs anxiety (over-planning) |
In Claude Code, you can install directly from the marketplace:
/plugin marketplace add supratikpm/atomic-habits-skills
/plugin install atomic-habitsOr clone it manually:
git clone https://github.com/supratikpm/atomic-habits-skills.git ~/.claude/plugins/atomic-habitsThen in Claude Code:
/plugin marketplace add ~/.claude/plugins/atomic-habits
/plugin install atomic-habitsThese skills are not just summarized quotes from the book. They have been rigorously translated into AI-native operational directives:
- Zero human analogies: We don't talk about getting 1% better at running; we talk about dropping cyclomatic complexity by 1%.
- Stack-agnostic placeholders: Every example uses
[target_module],[test_command], and[dependency_manifest]so they work universally across Node.js, Python, Rust, or Godot.
We believe the future of autonomous engineering is disciplined, agent-centric systems. If you have an idea for a new behavioral framework or want to refine an existing one, we welcome your contributions!
Read our Contributing Guidelines to get started.
This repository is a fan-created, open-source project born out of profound respect and love for Atomic Habits by James Clear.
We highly recommend purchasing and reading the original masterpiece at jamesclear.com/atomic-habits to fully understand the psychology behind these operations.
We do not claim ownership of the "Atomic Habits" trademark, the Four Laws of Behavior Change, or any of James Clear's intellectual property.
The sole purpose of this repository is to explore how these phenomenal human behavior principles can be translated into operational instructions for Artificial Intelligence Agents to build better, more resilient software systems.
To James Clear or his Publisher: If you feel this project infringes on your IP or you prefer these interpretations not to be hosted publicly, we will respectfully take this repository down immediately. Please simply open a thread in the Discussions tab of this repository, and we will comply without the need for formal DMCA reports. Until instructed otherwise through official channels, this project remains open-source for the developer community.