aksheyw

Claude Code Deep Review — Development skill for Claude Code

Development community

A 14-lens iterative deep review methodology for Claude Code.

How to install Claude Code Deep Review

This entry records only its repository, not the path inside it, so there is no exact command to give. Open aksheyw/claude-code-deep-review and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Claude Code Deep Review does

A 14-lens iterative deep review methodology for Claude Code. Found 14 production bugs (2 ship-stoppers) in its first use.

Alternatives in Development

Part 1 — Deep Research Prompt Builder — I'm going to help you create a research prompt for your project 2.1k ★
/gemini Consult — Engages in deep, iterative conversations with Gemini MCP for complex problem-solving 1.3k ★
Lens Reviewer — Use the @lens-reviewer agent to review the following code: $ARGUMENTS 523 ★

README

🔍 Deep Review: a 14-lens review loop for Claude Code

I review my own work fourteen times before I ship it, because one pass keeps missing things.

![requires](https://img.shields.io/badge/requires-Claude%20Code-D97757) ![lenses](https://img.shields.io/badge/lenses-14-0E9384) ![stops](https://img.shields.io/badge/stops-when%20a%20whole%20pass%20is%20clean-1B2A4A) ![install](https://img.shields.io/badge/install-2%20files-blue) ![dependencies](https://img.shields.io/badge/dependencies-none-brightgreen) ![license](https://img.shields.io/badge/license-MIT-green)

Claude Code is Anthropic's AI assistant for writing software, and this is one file you hand it. It reviews your work through fourteen different lenses, then keeps going until a whole pass finds nothing new.

The first time I ran it on a release it found **14 production bugs across 28 rounds**. Two would have broken every server call and every image upload on day one, and the review before had found neither.

Your pull request, test plan or design document goes through one pass of fourteen lenses: file coverage, the contract between app and server, security, state, config and nine more. If that pass finds anything new, all fourteen run again. When a pass finds nothing new, you ship.

Why I run fourteen lenses and not one

I was about to ship a release that had already been through a code review. Writing the test plan for it, I kept going back over the same work from a different angle each time, and new angles kept turning up what the earlier ones had missed. By the time I stopped there were fourteen production bugs, and two of them would have stopped the release.

Each angle catches a different *category* of problem, so I wrote the angles down as a skill so I'd never ship without running them, and this repo is that skill.

📋 The fourteen lenses, in the order they run