deft

Enforce rule-based guidance with USER.md, PROJECT.md, and language standards precedence.

28|13|Updated Dec 31, 2025
One-click install
npx skills add https://github.com/deftai/directive --skill deft-deftai
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deft
Source: https://github.com/deftai/directive/tree/main
Command: npx skills add https://github.com/deftai/directive --skill deft-deftai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Deft establishes a consistent, rule-based framework for AI-assisted development, reducing repetitive setup and ensuring aligned behavior across projects.

Core Features & Use Cases

  • Rule precedence: USER.md > PROJECT.md > language standards to ensure project-specific overrides are respected.
  • Lazy-loading context guidance: AI loads only the files needed for a task to minimize token usage and context.
  • Phase-driven workflows: Bootstrapping, specification generation, and CI-quality gates to enable reproducible AI-assisted work.
  • Standards-based collaboration: Clear conventions for naming, testing, and commit practices to streamline teamwork.

Quick Start

Tell your AI to apply Deft standards to your project and let it guide setup, coding, testing, and reviews.

Frequently Asked Questions about deft

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I enforce consistent project standards for AI-assisted development?

You can enforce consistent project standards for AI-assisted development by applying a rule-based framework that establishes a strict precedence hierarchy, ensuring project-specific overrides are always respected. Deft defines this hierarchy as USER.md overriding PROJECT.md, which then overrides language standards.

What is lazy-loading context guidance and how does it reduce token usage?

Lazy-loading context guidance is a mechanism where AI loads only the files required for a specific task. It minimizes token usage and context overhead by preventing the AI from processing unnecessary project files during development workflows.

How do I set up phase-driven workflows for reproducible AI-assisted coding?

To set up phase-driven workflows for AI-assisted coding, you apply a framework that structures work into bootstrapping, specification generation, and CI-quality gates. This enables reproducible results by enforcing strict operational directives and SKILL.md alignment.

Can I override language standards with project-specific rules in my development workflow?

Yes, you can override language standards with project-specific rules by defining them in PROJECT.md. The framework enforces a strict precedence where USER.md overrides PROJECT.md, and both override default language standards to ensure specific configurations are applied.

What's the best way to standardize naming, testing, and commit practices for AI collaboration?

The best way to standardize naming, testing, and commit practices for AI collaboration is to apply a standards-based framework that defines clear conventions for teamwork. This streamlines collaboration by ensuring all AI actions align with operational directives.

Why does my AI assistant ignore project-level configuration files during generation?

Your AI assistant may ignore project-level configuration files if a strict rule precedence model is not enforced. Applying a framework that prioritizes USER.md over PROJECT.md ensures project-specific overrides are recognized and respected during generation.