deft

Load hierarchical SKILL.md files with YAML frontmatter for AI coding assistants.

Updated Jan 8, 2026
One-click install
npx skills add https://github.com/sighthoundinc/hw-savings --skill deft
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deft
Source: https://github.com/sighthoundinc/hw-savings/tree/main/deft
Command: npx skills add https://github.com/sighthoundinc/hw-savings --skill deft

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Deft provides a hierarchical, lazy-loading SKILL.md framework that guides AI coding assistants with rule precedence and modular scope to ensure consistent, high-quality output.

Core Features & Use Cases

  • Rule precedence: personal user rules override project and language standards to tailor AI behavior.
  • Lazy loading: loads only the relevant deft files for a task to minimize context and costs.
  • Task-centric workflow: uses Taskfile and defined commands for initialization, project setup, specification generation, and QA.
  • Quality standards: enforces TDD/SDD, ≥85% test coverage, and conventional commits.
  • Language support: applies to Python, Go, TS, C++, and more with language-specific guidelines.

Quick Start

Start a new deft-enabled project, configure user/project, generate SPECIFICATION.md, and use the task runner for development.

Frequently Asked Questions about deft

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

FAQPage Schema
How do I manage context limits and reduce costs when using AI coding assistants?

Lazy loading minimizes AI context usage by loading only relevant SKILL.md files for a specific coding task, reducing operational costs and preventing context overflow.

How does rule precedence work when configuring AI coding framework standards?

Rule precedence in this AI coding framework allows personal user rules to override project and language-specific standards, ensuring tailored AI behavior while maintaining baseline quality.

Can I use this hierarchical SKILL.md framework with Python, Go, and TypeScript projects?

Yes, the framework applies language-specific guidelines across Python, Go, TypeScript, and C++ projects, supporting task-centric workflows for initialization, specification generation, and QA.

What is the best way to enforce test coverage and conventional commits in AI-driven development?

Use a task-centric workflow with Taskfile to enforce TDD/SDD, ensure at least 85% test coverage, and mandate conventional commits for consistent, reproducible AI-driven development.

How do I start a new project with a lazy-loading AI coding framework?

Start a new project by configuring user and project settings, generate a SPECIFICATION.md file, and use the defined task runner commands for subsequent development and testing operations.

Why do I need YAML frontmatter in SKILL.md files for AI coding tasks?

YAML frontmatter in SKILL.md files enables hierarchical rule precedence and modular scope, allowing the AI assistant to parse and load operational context dynamically as needed.