sandeco/reversa

Transform legacy systems into executable specifications for AI coding agents

What it solves

Reversa solves the problem of "trapped knowledge" in legacy software systems. It prevents AI coding agents from breaking critical, undocumented business rules and architectural decisions by converting existing code into executable, traceable specifications (operational contracts) that agents can use to evolve the system safely.

How it works

Reversa is installed as a framework within a legacy project. It coordinates a team of specialized AI agents to analyze the codebase and generate specifications. The process follows several distinct pipelines:

  • Discovery Pipeline: A five-phase sequence (Reconnaissance, Excavation, Interpretation, Generation, and Review) that maps the system, analyzes modules, extracts business rules, synthesizes architecture, and writes specifications.
  • Forward Cycle: A bridge that takes these specifications and evolves the system into running code through a sequence of requirements, planning, and auditing.
  • Migration Pipeline: Converts legacy specifications into a rebuild plan for a modern technology stack.
  • Greenfield Pipeline: For new projects, it moves from a raw idea to a full Software Design Document (SDD) specification.

All agents operate under a strict immutability rule: they never modify or delete existing project files, writing only to specific output folders (.reversa/ and _reversa_sdd/).

Who it’s for

Software engineers and architects working with legacy systems who want to use AI agents to maintain, evolve, or migrate their software without risking regressions in undocumented logic.

Highlights

  • Non-destructive analysis: Guaranteed immutability of the original source code.
  • Multi-agent orchestration: Uses specialized teams for discovery, ideation, bug tracking, and migration.
  • Causal traceability: Connects defects and changes back to the original specifications.
  • Operational contracts: Produces specifications designed for AI agents rather than just human-readable documentation.
  • Static documentation site: Can render extracted knowledge into a self-contained HTML mini-site with 3D architecture maps and quantitative metrics.

Related

  • Project
    Dryxio/auto-re-agent

    An AI-powered reverse-engineering agent that uses Ghidra and LLMs to reconstruct and validate C/C++ functions from compiled binaries through an autonomous reverser-checker pipeline.

  • Project
    morluto/rea

    An agent-driven reverse engineering framework that allows AI agents to investigate app behavior and native binaries using tools like Hopper and Ghidra to explain and recreate software features.

  • Project
    Jwuthri/Tracely-ai

    A trace-native CI/CD platform for AI agents that converts production failures into automated regression tests to block regressions in pull requests.

  • Project
    WhiteNightShadow/hello_js_reverse_skill

    A comprehensive toolkit and AI agent skill for Web reverse engineering, designed to restore API signatures and bypass anti-bot protections using an anti-detection browser and dual-language implementation templates.

  • Project
    lingbol088-spec/reverse-flow-skill

    A reverse engineering skill for AI agents that systematizes the process of analyzing binaries and CTF challenges through a structured analysis-to-report workflow.