visual-stack CLAUDE.md

Repository instructions for Visual Stack, a coding-agent plugin for interactive wireframing and reviewing user interfaces in a browser. Users comment on a wireframe or running app, and the agent applies the feedback to publish a new version.

In plain words
What is it for?
For building, reviewing, and updating interactive wireframes or app interfaces through browser comments, while maintaining the plugin repository and its host integrations.
Why use it?
It gives the coding agent the project’s commands, architecture, and rules so feedback can be handled consistently. It also explains how the plugin is packaged for different agent hosts.

Instructions file

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/cavalry-collective/visual-stack/claude-md
Clone the repo
git clone --depth 1 https://github.com/Cavalry-Collective/visual-stack
Per session 7,138 This file is loaded in full into every session.
When invoked 7,138 The same file — it is already loaded in full.
Security scan C 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.07138 $0.07138
Opus 5 $0.03569 $0.03569
Sonnet 5 $0.01428 $0.01428
Haiku 4.5 $0.00714 $0.00714

Measured 2d ago against content hash 75a2f5d81990, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

visual-stack CLAUDE.md scanned grade C with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

rm -rf $SANDBOX
CLAUDE.md · 553 lines

How it starts

The opening of the file, as written. The whole thing — 553 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

What this is

Visual Stack — a coding-agent plugin for interactive wireframing and UI review. The user comments directly on a wireframe (or a running app) in a browser workspace; the agent applies the feedback and publishes the next version into the same workspace. The repo is a Claude Code plugin marketplace (.claude-plugin/marketplace.json) with one plugin, plugins/vstack/, which also ships a Codex manifest (.codex-plugin/) and a Grok host adapter (skills/review/hosts/grok.md).

This repo is the thing people install; nothing is installed into it. No .grok/skills/, no .claude/skills/, no vendored copy of our own plugin — a host that discovers skills from a project directory gets instructions in its host adapter, not a checked-in skill directory that then has to be kept in sync.

.claude/commands/ is the one exception, and it holds maintainer tooling only: commands for working on this repo, which ship to nobody and mirror no part of the plugin. /ship is the only one. A command there states no rule of its own — it points at the section of this file that owns the rule.

Plain Node ≥ 18 ES modules, standard library only. There is no package.json, build step, bundler, or linter. (node_modules/ at the root appears only when recording the README demo, which installs playwright-core.)

Commands

Tests are standalone Node scripts — run them directly, one file per suite:

node plugins/vstack/skills/review/tests/review-lifecycle.mjs   # end-to-end review server round-trip
node plugins/vstack/skills/review/tests/host-profiles.mjs      # host profiles conform to host.schema.json
node plugins/vstack/skills/review/tests/workdir.mjs            # .vstack/local working-dir resolution
node plugins/vstack/skills/review/tests/round-gate.mjs         # `unanswered` and the Stop hook that runs it
node plugins/vstack/skills/review/tests/update-check.mjs       # per-host update detection and the banner it produces
node plugins/vstack/skills/review/tests/design-tokens.mjs      # the shell's palette still matches design/tokens.css

Read the full file on GitHub · 553 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 553 lines · 7,138 tokens per session scan C 75a2f5d81990

Subscribe to this mod's changes

visual-stack CLAUDE.md is an instructions file published in the GitHub repository Cavalry-Collective/visual-stack (11 stars, last pushed 13d ago), licensed MIT. It adds 7,138 tokens to every session, about $0.0357 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.