reversa-coding

Executes actions.md plans into code while generating legacy-impact and regression-watch audit artifacts.

Updated Jun 26, 2026
One-click install
npx skills add https://github.com/RLuf/fzagent --skill reversa-coding-rluf
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: reversa-coding
Source: https://github.com/RLuf/fzagent/tree/main/.agents/skills/reversa-coding
Command: npx skills add https://github.com/RLuf/fzagent --skill reversa-coding-rluf

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? It turns a structured implementation plan (actions.md) into actual code changes while preserving traceability against a reverse-engineered legacy baseline, so every modification to legacy behavior is documented and watchable for regressions. ## Core Features & Use Cases - Phased plan execution: Runs pending actions phase by phase (Preparação, Testes, Núcleo, Integração, Polimento), respecting dependencies and parallel markers, updating checkboxes and appending to progress.jsonl. - Legacy impact auditing: Generates legacy-impact.md mapping touched files to components from the reverse extraction, classifying impact type and severity, and listing preserved versus modified rules. - Regression watch tracking: Creates regression-watch.md with stable watch item IDs for modified or removed rules, feeding future re-extraction cycles. - Use Case: After running the reverse pipeline and /reversa-to-do on a legacy codebase, invoke this skill to implement the planned feature while automatically producing an audit trail of what legacy behavior changed and what must keep holding true. ## Quick Start Type /reversa-coding to execute the pending actions in the active feature plan and generate the legacy impact and regression watch reports.

Frequently Asked Questions about reversa-coding

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

FAQPage Schema
How do I execute an actions.md implementation plan with an AI agent?▼

Invoke /reversa-coding after generating actions.md with /reversa-to-do. The skill runs pending actions phase by phase, checks off completed items, logs each action to progress.jsonl, and stops the phase if any action fails.

What is the reversa framework forward coding phase?▼

It is the final stage of the reversa cycle where the planned actions become real code. It requires the reverse extraction output (_reversa_sdd) as an anchor so legacy-impact.md and regression-watch.md can trace every change back to documented legacy rules.

Why does reversa-coding abort when _reversa_sdd is missing?▼

The skill strictly requires _reversa_sdd containing architecture.md and domain.md from the reverse pipeline. Without them, the legacy impact and regression watch artifacts lose their anchor, so the skill aborts without writing anything to disk.

Can I run only part of the plan, like specific task IDs?▼

Yes. Pass a free argument indicating a phase or ID interval, such as a single phase name or a range like T001-T005, and execution is restricted to that scope. Otherwise all incomplete actions run in order.

What happens when an action fails during execution?▼

The failed action keeps its unchecked status in actions.md, a failed status line is appended to progress.jsonl, and the current phase stops. The skill reports the failure to the user instead of continuing silently.

Which AI coding assistants support this skill?▼

The skill declares compatibility with Claude Code, Codex, Cursor, Gemini CLI, and other agents supporting the Agent Skills format, since it is a standard SKILL.md-based unit with no external script dependencies.