prd generator agent

Writing a Product Requirements Document (PRD) often feels like a chore, leading to vague specs and scope creep. This skill replaces the manual writing process with a conversational, bilingual interview that extracts the right details from your brain or your existing code to generate production-ready documentation.

What it does

• Automated context detection scans your existing repository, package.json, and source code to pre-fill requirements and tech stack details.
• Bilingual interview flow conducts a 16-question discovery process in English or Indonesian to nail down user stories, priority buckets, and metrics.
• Strict validation runs an automated check to ensure no sections are missing, placeholders are removed, and every user story has clear acceptance criteria.
• Multi-format exports generates not just a Markdown PRD, but also a Mermaid roadmap, a JSON task list for your issue tracker, and a .cursorrules file.
• Context-aware revision handles updates to existing PRDs without losing historical context or drifting from the current codebase state.

How it works

1. Initialize by providing a project name, a raw idea, or a path to an existing repository.
2. Interview through four structured rounds of questions, skipping anything the skill already discovered from your files.
3. Verify as the skill runs a validation script to catch technical gaps or missing priority levels.
4. Export the final assets, including the PRD, roadmap, and project-specific AI rules directly to your workspace.

Frameworks & tools

Works with any tech stack by reading standard configuration files like package.json or pyproject.toml. It utilizes node.js for validation and export scripts, Mermaid for roadmap rendering, and standard Markdown for documentation.

Why this beats prompting it yourself

Generic prompts often hallucinate features you don't need or miss critical acceptance criteria. This skill uses a structured, multi-phase procedure with a dedicated validator to ensure the output is actually useful for development, not just a wall of text.

Use cases

• Drafting a full PRD for a new app idea from a blank slate.
• Generating missing documentation for a legacy repo by analyzing existing code.
• Breaking down a high-level feature request into validated user stories and JSON tasks.
• Creating a roadmap and .cursorrules to help AI coding agents understand your project's constraints.

Known limitations

Requires a Node.js environment to run the internal validation and export scripts. The interview rounds are capped at 4 questions per turn to maintain focus.