AI Coding Skill Dojo

Turn your coding preferences into model-agnostic reusable Skill packs and track results

Problem

When coding with AI, the code often runs but is messy—redundant null checks, inconsistent style. Personal prompts and context live scattered across different tools, vanish when switching models, and never evolve with usage.

Solution

A local-first Skill manager: users write maintainability rules and project conventions as versioned mini Skill files, export them to formats recognized by different models and editors, and record adoption and rework rates of each generation to iterate their own 'coding brain'.

Users

Indie developers and tech leads who heavily use Cursor/Claude Code and care about code quality

MVP

CRUD and version history for Skill files; export to common AI editor formats; simple quality snapshots of generated code (cyclomatic complexity, duplication); one-click comparison across Skill versions; local storage with optional Git sync

Revenue

One-time purchase or low annual subscription for individuals; per-seat team plan with shared Skill library and stats dashboard

Why now

AI coding tools are mainstream but maintainable code remains a common pain; users already discuss capturing skills and agents; existing tools generate but don't accumulate—a solo dev's lightweight layer fits the gap