# browser-verifier

browser-verifier takes a client-side finding backed by curl evidence and settles whether the payload fires in a real browser. Useful before an XSS report goes out.

- skill · `type=skill` · Claude Code, Cursor, Codex, GitHub Copilot, Gemini CLI, Windsurf (Devin Desktop), Claude.ai · ★ 1298 repo stars
- Author: H-mmer
- License: not stated · policy=link-only · description only; the file stays with the author
- Source: https://github.com/H-mmer/pentest-agents/tree/HEAD/providers/cursor/.cursor/skills/agent-browser-verifier/SKILL.md
- Updated: 2026-06-12
- Security check: automated, 2026-11-16 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/skills/browser-verifier

## What it does

The skill loads a client-side finding so one question gets answered: does the payload actually run in a browser? XSS, DOM sinks, `postMessage` and prototype pollution all go through it. A response body that echoes your input proves very little, because CSP, framework auto-escaping, a SPA that renders differently or a WAF challenge aimed at the browser request can each stop execution.

A `curl` header check comes first, to see whether Cloudflare sits in front of the target. Then the camofox stealth browser starts with `../../tools/camofox_ctl.sh start`, and the payload URL loads in a tab driven over `http://localhost:9377/tabs`. Detection does not lean on `alert()`: dialog hooks, a DOM marker such as `XSS-FIRED`, or an out-of-band callback. With a strict CSP, preload links and DNS prefetch carry the exfiltration instead.

Three verdicts come back: `BROWSER_CONFIRMED`, `BROWSER_REJECTED`, and `BROWSER_PARTIAL` when the code runs but data cannot leave, which lowers severity. A screenshot lands in `evidence/`, and the JSON output points back at the original finding. A finding is not dead because `alert(1)` stayed silent: the later detection tiers have to come back empty first.

## Install

### Claude Code (native platform)

1. **Skills CLI (Vercel)**
```bash
npx skills add H-mmer/pentest-agents --skill browser-verifier -a claude-code
```

2. **Manually: put the files in place**
- File: `~/.claude/skills/agent-browser-verifier/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/agent-browser-verifier/

### Codex

1. **Skills CLI (Vercel)**
```bash
npx skills add H-mmer/pentest-agents --skill browser-verifier -a codex
```

2. **Manually: put the files in place**
- File: `.agents/skills/agent-browser-verifier/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

1. **Skills CLI (Vercel)**
```bash
npx skills add H-mmer/pentest-agents --skill browser-verifier -a cursor
```

2. **Manually: put the files in place**
- File: `.cursor/skills/agent-browser-verifier/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install H-mmer/pentest-agents providers/cursor/.cursor/skills/agent-browser-verifier
```

2. **Manually: put the files in place**
- File: `.github/skills/agent-browser-verifier/SKILL.md`

### Gemini CLI

1. **Skills CLI (Vercel)**
```bash
npx skills add H-mmer/pentest-agents --skill browser-verifier -a gemini-cli
```

2. **Platform CLI**
```bash
gemini skills install https://github.com/H-mmer/pentest-agents --path providers/cursor/.cursor/skills/agent-browser-verifier
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/agent-browser-verifier/SKILL.md` — The legacy .windsurf/skills/ path still works

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/H-mmer/pentest-agents/tree/HEAD/providers/cursor/.cursor/skills/agent-browser-verifier/SKILL.md

## When to use it
- rule out an XSS finding that reflects but never executes
- back a confirmed execution with a DOM snapshot and a screenshot
- work out why a payload stays silent: CSP, escaping, WAF, wrong context
- downgrade a finding that runs but cannot exfiltrate
- keep the verdict and the finding reference in the JSON output

## Example prompts
- Check the client-side finding from the report in a real browser: does the payload execute or not?
- Run `/hunt` on this target and hand the client-side findings to browser-verifier.
- The payload reflects in `curl` and looks like XSS. Prove or disprove it in a browser and save the screenshot under `evidence/`.

## Similar
- [hunt-xss](https://affpapa.org/en/nevibecode/skills/hunt-xss.md) — skill · ★ 16193 repo stars
- [fix-security-vulnerabilities-with-strix](https://affpapa.org/en/nevibecode/skills/fix-security-vulnerabilities-with-strix.md) — skill · 20300 installs
- [injection-checking](https://affpapa.org/en/nevibecode/skills/injection-checking.md) — skill · 4100 installs
- [pentesting-checklist-security-assessment](https://affpapa.org/en/nevibecode/skills/pentesting-checklist-security-assessment.md) — skill · 180 installs
- [fix-verifier](https://affpapa.org/en/nevibecode/agents/fix-verifier.md) — subagent · ★ 2482 repo stars
- [xss-html-injection](https://affpapa.org/en/nevibecode/skills/xss-html-injection.md) — skill · ★ 4719 repo stars

From the same set: [h-mmer/pentest-agents](https://affpapa.org/en/nevibecode/by/h-mmer/pentest-agents.md) — 43 cards

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
