coding-assistant

Classifies code and config changes as destructive/non-destructive and coordinates Claude review sequences with verification reporting.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/nowonbun/nowonbun-harness --skill coding-assistant-nowonbun
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: coding-assistant
Source: https://github.com/nowonbun/nowonbun-harness/tree/main/codex-skills/action-management_coding-assistant
Command: npx skills add https://github.com/nowonbun/nowonbun-harness --skill coding-assistant-nowonbun

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ensures coding, configuration, and developer-document changes are risk-classified, correctly reviewed, and verifiably reported so that destructive changes are gated and verification evidence is captured.

Core Features & Use Cases

  • Change Classification: Distinguishes destructive vs non-destructive changes and documents approval requirements.
  • Coordinated Review Sequences: Orchestrates plan-review, source-review, and result-review steps, leveraging Codex for implementation and Claude for independent reviews with defined fallbacks.
  • Verification & Reporting: Requires explicit verification commands, evidence capture, and final reports that separate verified facts from assumptions; useful for release gating, schema/API changes, and config updates.

Quick Start

Use the coding-assistant skill to classify a proposed change as destructive or non-destructive, generate the required review sequence, and produce a final verification report.

Frequently Asked Questions about coding-assistant

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

FAQPage Schema
How do I classify code changes as destructive or non-destructive for release governance?

Classify code changes as destructive or non-destructive to enforce approval gating and document specific requirements for release governance. This risk classification ensures destructive modifications are properly gated before execution and review.

What is a coordinated code review sequence for source changes and implementation?

A coordinated code review sequence orchestrates plan-review, source-review, and result-review steps for implementation changes. It leverages Codex for implementation tasks and Claude for independent reviews with defined fallbacks.

How do I generate a verification report for schema and API configuration updates?

Generate a verification report for schema and API configuration updates by requiring explicit verification commands and capturing evidence. The final report separates verified facts from assumptions and highlights residual risks.

Can I use Claude for independent code review if Codex handles the implementation?

Yes, you can use Claude for independent code review while Codex handles implementation. The workflow applies Claude review sequences across plan, source, and result phases, including defined fallback handling if a review step fails.

What are the limitations of automated risk classification for developer-document changes?

Automated risk classification for developer-document changes requires explicit verification commands and may still leave residual risks. Limitations include the need to separate verified facts from assumptions and manage approval gating manually for destructive updates.

When do I need approval gating for configuration changes?

You need approval gating for configuration changes when they are classified as destructive. The classification process documents specific approval requirements and enforces gating to ensure risk-aware execution before any review sequence proceeds.