MySpec
An AI agent that bridges the gap between vibe coding and spec-driven development.
Core trick
An agent asks clarifying questions and writes requirements, design, and task files that vibe-coding tools consume as the spec.
Description
An AI agent that bridges the gap between vibe coding and spec-driven development. It turns your idea into executable documentation. It asks the right questions to produce a “Spec Bundle” — clear, detailed Requirements, Design, and Task files — that can be fed directly into Gemini CLI, Google AI Studio, or other vibe-coding tools to build with far more precision so your coding agent works from a clean source of truth instead of guesswork.
README excerpt
# MySpec Monorepo (Prototype) > **Purpose**: A simplified prototype for new team members to understand the architecture and request flow of the MySpec system. This is NOT production code — it's a concept-level walkthrough. ## Built with Codex & GPT-5.6 Codex, powered by GPT-5.6, was used as the development collaborator for this prototype. It helped turn the original MySpec codebase and project goals into a smaller, easier-to-follow monorepo by: - exploring the existing architecture and tracing the end-to-end request flow; - scaffolding the React, Cloudflare Worker, Durable Object, WebSocket, and Bun service boundaries; - translating the multi-agent workflow into focused TypeScript modules and system prompts; and - producing the walkthrough documentation that helps new contributors understand the system quickly. GPT-5.6's long-context reasoning was especially useful for keeping shared protocols aligned across the webapp and AI-agent service while preserving the relationships between sessions, streaming responses, orchestration, and specialist agents. > **Development vs. runtime:** Codex and GPT-5.6 were used to build and document this prototype. The prototype's example runtime agent integration uses the Anthropic Claude SDK, as shown in the architecture below. ## What Is MySpec? MySpec is an AI-powered spec generation tool. Users describe their product idea in a chat interface, and a team of AI agents collaborates to produce technical specifications. ## Project Struct