What are the context levels for?

Context levels (min, standard, full) allow you to control how much project information is loaded into the AI's memory, optimizing performance and reducing token costs.

How does this skill help with long coding sessions?

It creates persistent artifacts in a .specs/ folder, allowing the AI to resume exactly where it left off with full context in a new session, avoiding 'memory' loss.

Does it automate task tracking?

Yes, it utilizes a tasks.yaml file to track progress, dependencies, and completion criteria, providing a clear roadmap for the AI agent to follow.

What is Specification-Driven Development for AI?

It is a methodology where AI development is guided by detailed, formal specifications and technical plans to ensure accuracy, consistency, and alignment with user requirements.

Can I use this for existing projects?

Yes, it includes specific support for 'brownfield' development using delta files to track modifications, additions, and removals in existing codebases.

Spec-Driven Development

Name: Spec-Driven Development
Author: ThilinaTLM

byThilinaTLM

0•

Productivity & Workflow

Implements a structured, multi-session workflow for AI agents using specification-driven development patterns to manage complex software features.

Spec-Driven Development is a robust framework designed to help Claude handle complex, multi-file software engineering tasks that exceed single-session context limits. By breaking down projects into formalized specifications, technical plans, and YAML-based task lists, it provides AI agents with a persistent source of truth and clear implementation guardrails. Whether you are building new features from scratch or modifying 'brownfield' legacy code, this skill ensures continuity, reduces hallucinations, and maintains high standards for technical requirements by utilizing token-optimized artifacts and structured workflow phases.

Key Features

01Multi-level context loading (min, standard, full) to optimize token usage across sessions

02Integrated CLI for initializing, planning, and archiving development specifications

03Structured specification management via a dedicated .specs/ directory hierarchy

04Brownfield modification support using delta-based requirement tracking for existing codebases

05Automated task breakdown and progress tracking with YAML-based checkpoints

060 GitHub stars

Use Cases

01Managing large feature implementations that span multiple files and require several coding sessions

02Standardizing AI agent behavior during complex refactoring or legacy code modifications

03Ensuring cross-session continuity and persistent requirement tracking for AI agents

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add thilinatlm/claude-plugins spec-driven-dev

For use in Claude.ai and ChatGPT

Download Skill