reversa-reviewer

Review engineering reverse-specifications for consistency and reclassify evidence confidence.

Updated May 5, 2026
One-click install
npx skills add https://github.com/AlexandrePontesjr/calculadora-Juridica --skill reversa-reviewer-alexandrepontesjr
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: reversa-reviewer
Source: https://github.com/AlexandrePontesjr/calculadora-Juridica/tree/main/.agents/skills/reversa-reviewer
Command: npx skills add https://github.com/AlexandrePontesjr/calculadora-Juridica --skill reversa-reviewer-alexandrepontesjr

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you catch inconsistencies, incorrect confidence classifications, and missing requirements in engineering reverse-spec outputs before they are used for implementation.

Core Features & Use Cases

  • Critical spec review: Questions and tests generated specifications to improve correctness and internal coherence.
  • Confidence reclassification: Reclassifies claims using confidence rules (confirmed/inferred/gap) and enforces evidence-based upgrades/downgrades.
  • Human validation routing: Produces targeted questions for human stakeholders to resolve remaining 🔴 gaps.
  • Optional cross-review: When enabled, it can delegate an independent review to an external engine (Codex) for a second opinion.

Quick Start

Use the reversa-reviewer skill after a reversa-writer run, ensuring your state.json and output_folder specs are present, so it can read units, apply confidence rules, and generate the final confidence report and validation questions.

Frequently Asked Questions about reversa-reviewer

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

FAQPage Schema
How do I validate reverse-engineered specifications for consistency?

Spec review detects missing behaviors, contradictions, and weak confirmations by questioning assumptions and reclassifying evidence confidence across engineering documentation artifacts before implementation begins.

What is the best way to check confidence classifications in engineering specs?

The best way to check confidence classifications is to apply reference confidence rules to reclassify evidence as confirmed, inferred, or gap. This enforces evidence-based upgrades and downgrades across generated reverse-specifications.

How do I generate human validation questions for spec gaps?

To generate human validation questions, route remaining evidence gaps to human stakeholders by producing targeted questions outputs. This resolves missing requirements and validates uncertain claims identified during the spec review.

Can I use an external engine for cross-review of generated specs?

Yes, you can enable optional cross-review to delegate an independent review to an external engine like Codex for a second opinion on your reverse-specifications.

Do I need state.json and config.toml to review reverse-specifications?

Yes, reviewing reverse-specifications requires reading state.json and config.toml to determine specs granularity. These files provide the current state and configuration needed to apply confidence rules and generate reports.