reversa-coding

Convert phase-based action plans into code with audit artifacts.

1|Updated Dec 20, 2025
One-click install
npx skills add https://github.com/Rdinda/FMA_Pontos --skill reversa-coding-rdinda
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: reversa-coding
Source: https://github.com/Rdinda/FMA_Pontos/tree/main/.agents/skills/reversa-coding
Command: npx skills add https://github.com/Rdinda/FMA_Pontos --skill reversa-coding-rdinda

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It solves the problem of transforming a forward “to-do” action plan into real, phase-by-phase implementation code while keeping legacy consistency and producing audit artifacts.

Core Features & Use Cases

  • Executes actions into code: Converts feature-dir/actions.md unfinished items into working implementation, updating each checkbox as it succeeds.
  • Maintains execution traceability: Appends immutable status entries to feature-dir/progress.jsonl, including timestamps and files touched.
  • Creates legacy audit trails: Generates legacy-impact.md (what changed in the legacy vs what remains true) and regression-watch.md (what must be re-validated to prevent regressions), tied to _reversa_sdd/domain.md and _reversa_sdd/architecture.md.
  • Applies phase scoping & dependency handling: Runs actions by phase (Preparação, Testes, Núcleo, Integração, Polimento) and respects independent action groups and flagged sub-conjuntos.
  • Guardrails via hard preconditions: Refuses to run unless the reversa extraction exists and includes required legacy artifacts (_reversa_sdd/architecture.md and _reversa_sdd/domain.md), preventing orphan audit outputs.

Quick Start

Run /reversa-coding to execute the current active feature’s actions into code and generate legacy-impact.md and regression-watch.md from the reversa-extracted legacy context.

Frequently Asked Questions about reversa-coding

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

FAQPage Schema
How do I turn a phase-based actions plan into working code with regression safeguards?

To turn an actions plan into working code, execute unfinished items in feature-dir/actions.md phase by phase. The process updates progress.jsonl append-only and generates legacy-impact.md and regression-watch.md for traceable regression prevention.

How does repo automation handle execution traceability during code generation?

Execution traceability is maintained by appending immutable status entries to feature-dir/progress.jsonl. These entries include timestamps and specific files touched, ensuring a complete audit trail for the automated coding workflow.

Do I need a prior reversa extraction directory to execute feature actions into code?

Yes, you need a prior reversa extraction directory named _reversa_sdd/ containing architecture.md and domain.md. Hard preconditions refuse to run without these required legacy artifacts, preventing orphan audit outputs.

Can I run a scoped subset of actions instead of the entire feature plan?

Yes, you can optionally execute an ID-scoped subset of the feature plan. The workflow applies phase scoping and respects independent action groups and flagged sub-conjuntos during code generation.

What's the best way to prevent legacy regressions when implementing new code changes?

To prevent legacy regressions, generate legacy-impact.md and regression-watch.md tied to domain.md and architecture.md. These audit artifacts define what changed in the legacy system and what must be re-validated.

Why does workflow execution fail when starting a new feature coding phase?

Workflow execution fails if the required _reversa_sdd/ extraction directory is missing or lacks architecture.md and domain.md. Guardrails enforce these hard preconditions to prevent orphan audit outputs.