ClaudeCodeMod

All shelves / Mods / Review

visual-review

By dougcunha

visual-review is a Claude Code review mod by dougcunha.

Runs /visual-review so a subagent models your branch's design, shows it as a diagram in the browser, and sends your comments back to the session in one step.

Draws

Install

The marketplace name is claude-mods.

In Claude Code
/plugin install visual-review --marketplace dougcunha/claude-mods
Open the repo

It runs inside the Claude Code process. The side panel is only what the author's README names. Read the repo before you install, or run claude plugin validate on the plugin folder.

Version
0.3.0
License
MIT
Repo stars
4
Last change
2026-10-23

What it can reach

  • Draws in the session
  • Calls another program on this machine

Answer y to add the marketplace and choose a scope; with it added, /plugin install visual-review@claude-mods. It needs a Git repository. The interactive page needs Node.js on the PATH; without it the page is static, and Copy review puts the review on the clipboard. Each run starts one subagent, Sonnet by default, set with analysisModel in /config. Tested on Windows 11 with Chrome and Chromium; the browser launch on macOS and Linux is not tested yet.

The marketplace, claude-mods, holds dougcunha's four mods: session-info, topic-guard, feature-progress, and visual-review. Seeing the design of a change before you merge it is why this page files it under review. crit-tui opens a pane where you comment on lines of the diff and sends the review to Claude in one go. leaf has the agent build a browser page you comment on and edit. This one draws the branch's classes, modules, relations, and database tables against a base, with line counts read from Git rather than from the model, coupling figures, and cohesion notes, in Overview, Diagram, and List views. Claude answers your questions and proposes a plan for change requests, and changes no file before you confirm. In the source, the session shows a status line while the analysis runs and a toast with the page's link.

Demo

The What it does section of the visual-review README: /visual-review collects the branch's changes against a base; the mod sends one prompt, and a subagent sends a design model to the mod's submit_model tool; the mod checks that every changed file is covered, reads the line changes from Git, and serves the page from a small Node.js server on 127.0.0.1, or writes a static page when node is missing.
Screenshot of the author's GitHub page.

What it hooks into

  • session.startWhen a session starts
  • command.runWhen you run its slash command
  • tool.callAround a tool call, where it can hold one
  • turn.startWhen Claude starts a turn
  • agent.spawnWhen a subagent starts
  • turn.completeWhen a turn ends
  • session.endWhen a session ends

Read from the on() calls in its hook modules. Open its source under Files to check what each one does.

Files

The author's README and source, 4 files. They load when you open them.

Facts

Author
dougcunha
Category
Review
Marketplace
claude-mods
Repo
dougcunha/claude-mods
Folder
visual-review
Version
0.3.0
License
MIT
Repo stars
4
Last change
2026-10-23
Language
TypeScript
Files
33

Other mods in Review

archpanefacumarcetOpens a live architecture diagram in a side pane that Claude draws and keeps current as you talk, with subdiagrams, build status by color, and clickable boxes.file-treeha-ptt0601Draws an IDE-style file tree beside the conversation that lights up what Claude reads, edits, and commits, with git status, search, and a file preview.debuggerHula-Hoop-AIHolds a commandOpens a step debugger pane for the agent loop: pause at a prompt, model request, response, or tool call, edit it or skip the call, then step, play, or stop.frontend-skills-modsmalinskibeniaminOpens a Harness desk pane of check receipts, the last 24 Skill calls, and replayable edit patches, with a context and last-skill readout above the prompt.slop-filterNicoSKOOLHolds a commandScores each page of website copy Claude writes for AI tells and SEO cliches, in English and Spanish, refuses one that reads like AI, and shows the score.plain-englishnordscope-fiHolds a commandRefuses or asks before a Markdown write, commit message, or tracker issue that breaks its prose rules, quoting the passage, and holds a reply for a rewrite.cc-grammarrapitaShows the English of your draft in gray after a pause, translated or corrected, and one key takes it in. A message sent unseen gets a gray line of fixes.traceHula-Hoop-AIOpens a /trace pane of the skills that loaded this session and the web pages Claude fetched or saw in search results, ready to insert as citations.change-journaltheonly1meOpens a live page in your browser of every change Claude makes, each diff with its reason, plus a /changes pane and a count of files on the status line.decision-lenstheonly1meLists the decisions Claude made in each turn in a /why pane, with why and what it chose not to do, and keys to keep or stop a behavior in later turns.lookstheonly1meRedraws the transcript in one of four themes, Retro CRT, Punk, Synthwave, or Zen paper, with tidier tool rows, and optional one-line rows or centered replies.unsloptheonly1meFinds AI slop in each edit Claude makes, such as dashes, comment blocks, and tests that assert nothing, has Claude remove it, and counts it on the status line.
All 145 in Review