The Evolution of Spec-Driven Tooling
GitHub's open-source Spec-Kit helped popularize markdown-based specifications for AI coding. However, CLI-only templates require developers to manually write boilerplate specifications, manually debug acceptance criteria syntax, and manually copy files into IDE contexts.
MySpec takes a different approach: an automated Senior Architect AI engine that interviews you in 2 minutes, extracts hidden edge cases, formats requirements using EARS+ syntax, and streams the live bundle directly into Cursor, Claude Code, and Windsurf via MCP.
Comprehensive Feature Comparison
| Feature | GitHub Spec-Kit | MySpec |
|---|---|---|
| User Interface | Terminal CLI only | Interactive Web App + MCP Server |
| Spec Generation | Manual Markdown typing | 2-Min Socratic AI Senior Architect Interview |
| Requirements Linting | None (Freeform text) | Automated EARS+ Syntax Validation |
| Diagram Generation | Manual ASCII / Text | Live Mermaid.js System Diagrams |
| IDE Streaming | Manual file reading | Native Model Context Protocol (@myspec/mcp-server) |
| Brownfield Codebase Support | Limited | Automated Codebase Reverse-Engineering & Proposal Deltas |
60-Second Migration from Spec-Kit to MySpec
Already have specs written with Spec-Kit? Import your existing markdown files directly into MySpec with a single command:
# Import your local .specs into MySpec
npx @myspec/mcp-server import ./.specs Switch to MySpec
Generate your first 4-file spec bundle with zero CLI boilerplate.