spec-miner

Extracts specifications and requirements from existing codebases using code analysis.

Updated Mar 9, 2026
One-click install
npx skills add https://github.com/ArMaTeC/Redball --skill spec-miner-armatec
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: spec-miner
Source: https://github.com/ArMaTeC/Redball/tree/main/.devin/skills/spec-miner
Command: npx skills add https://github.com/ArMaTeC/Redball --skill spec-miner-armatec

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Legacy and inherited codebases often lack documentation, making it hard to understand what the system actually does. This Skill reverse-engineers specifications directly from source code, producing structured requirements documents grounded in real code evidence. ## Core Features & Use Cases - Codebase Exploration: Maps entry points, routes, models, auth, and integrations using Glob and Grep patterns. - EARS Requirements: Documents observed behavior in the Easy Approach to Requirements Syntax format with code locations. - Structured Specification Output: Generates a complete reverse-engineered spec covering architecture, functional and non-functional requirements, uncertainties, and recommendations. - Use Case: You inherit an undocumented NestJS API. Run this Skill to trace its endpoints, data models, and auth flows, then receive a specification document with EARS-formatted requirements and flagged open questions. ## Quick Start Analyze this codebase and generate a reverse-engineered specification document with observed requirements in EARS format.

Frequently Asked Questions about spec-miner

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I document a legacy codebase with no documentation?▼

Reverse-engineer the specification by exploring entry points, routes, models, and auth patterns with Glob and Grep, then document observed behavior in EARS format. The output is a structured spec covering architecture, requirements, and open questions.

How to generate API documentation from source code?▼

Trace route definitions and controllers using patterns like @Controller, router.get, or @app.route, then record endpoints, methods, request/response formats, and authentication requirements. Each observation includes its code location as evidence.

What is the EARS format for requirements?▼

EARS (Easy Approach to Requirements Syntax) structures requirements as patterns like 'When [trigger], the system shall [action]' for events or 'While [state], the system shall [action]' for states. It produces unambiguous, testable requirement statements.

Can reverse-engineered specs distinguish facts from assumptions?▼

Yes. Observations must be grounded in actual code evidence with file locations, while inferences and uncertainties are flagged in a dedicated section. Assumptions without code evidence are explicitly prohibited.

What are the limitations of code-based specification extraction?▼

It captures only what the code implements, not original intent or business rationale. Undocumented business rules, deployment context, and planned-but-unimplemented features must be flagged as questions for stakeholders.