custom-code-review-html

Render code review findings as a self-contained triageable HTML report with markdown hand-off.

1|Updated Apr 3, 2021
One-click install
npx skills add https://github.com/NaoyaMiyagawa/dotfiles --skill custom-code-review-html-naoyamiyagawa
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: custom-code-review-html
Source: https://github.com/NaoyaMiyagawa/dotfiles/tree/main/.ai/skills/custom-code-review-html
Command: npx skills add https://github.com/NaoyaMiyagawa/dotfiles --skill custom-code-review-html-naoyamiyagawa

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and assets (resource) components.

What problem does it solve? Terminal code reviews are read once and scrolled past, making it hard for reviewers to work through findings one at a time or hand accepted fixes to an implementing session. This Skill turns a diff review into a persistent, interactive HTML page where each finding can be triaged and exported. ## Core Features & Use Cases - Structured review workflow: Enforces a blind pass before reading the PR description, an intent pass against the plan, and proof-by-running for inferred findings, with verified file:line citations. - Triageable HTML report: Builds a self-contained page from a template with ranked findings, before/after code panes, accept/reject triage persisted in localStorage, optional EN/JA language switching, and a markdown feedback generator for the fixing session. - Pre-publish validation: Ships a Python checker that catches unbalanced tags, unpaired language blocks, font-relative widths, and unparseable inline JavaScript before the report is published. - Use Case: After reviewing a large pull request, generate a single HTML file the reviewer opens in a browser, accepts or rejects each finding, then copies the generated markdown of accepted items straight into the session that will implement the fixes. ## Quick Start Review the current pull request diff and produce an interactive HTML review report I can triage in my browser.

Frequently Asked Questions about custom-code-review-html

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

FAQPage Schema
How do I turn a code review into an HTML report?

Review the diff using the blind-pass then intent-pass workflow, rank findings by how much logic rests on them, then fill in the provided HTML template with one article block per finding. Run the included check-report.py script to validate the page before publishing it as an artifact or local file.

How do I review a GitHub pull request diff before writing findings?

Fetch the diff with gh pr diff and the metadata with gh pr view, then review the diff before reading the PR description so the plan cannot excuse weak implementation. Afterwards read the description to keep or dissolve each blind-pass finding with a stated reason.

Can the review report support both English and Japanese?

Yes, the template ships both languages as real lang="en" and lang="ja" markup with one hidden by CSS. Layout drift is prevented by using rem-based widths, a single type scale for both scripts, and scroll anchoring across the language switch.

Why does the report checker fail on font-relative widths?

The checker rejects ch or em values in width and max-width declarations because a font-relative measure resizes the whole column when the font stack changes on language switch. Use rem units instead so the layout stays stable between English and Japanese.

What are the limitations of the HTML review report format?

The page must be fully self-contained with inline CSS and JS, no CDNs or web fonts, and no document skeleton tags since the artifact host wraps the file. Triage state lives in browser localStorage keyed by a unique slug, so it is not shared across machines or browsers.