spec_driven_develop

Overview

This skill provides a comprehensive, seven-phase methodology for handling complex software engineering tasks such as major refactors, rewrites, and architecture migrations. It integrates deep project analysis, S.U.P.E.R health evaluation, GitHub-native issue and milestone tracking, parallel lane execution with worktrees, and closed-loop adaptive control based on execution telemetry and drift scoring. Designed with zero third-party runtime dependencies, the repository relies on pure Markdown references and lightweight helper scripts to guide AI agents effectively across multiple platforms.

Capabilities

• ▸Seven-phase spec-driven development workflow (Phases 0-6)
• ▸GitHub-native integration for issues, milestones, labels, and project boards
• ▸Adaptive control loop with telemetry tracking and drift-based self-correction
• ▸Orchestrator-centric tiered execution with optional parallel lanes and worktrees
• ▸Structured deep discussion and brainstorming workflow
• ▸Findings-first code review workflow focusing on bugs and regressions

Best for

Executing large-scale project rewrites or migrations into new languages/frameworks, Breaking down complex architectural changes into manageable, trackable phases and tasks, Automating GitHub issue creation, milestones, and batched pull requests through AI agents, Conducting structured technical analysis and deep discussions for complex problem solving, Performing findings-first code reviews on uncommitted changes, date ranges, or PR diffs

Works with

Security considerations

The skill runs locally using standard developer CLI tools (git, bash, gh) and executes instructions provided in Markdown files. Users should ensure they trust the repository and agents running these workflows, as they interact directly with local codebases and remote repositories.

Categories