A browser-control tool for opening pages, clicking, filling in forms, taking screenshots, and running JavaScript in Chrome or Chromium.

Use it to navigate websites, operate buttons and forms, inspect changing page state, and capture screenshots.

It handles web tasks that need interaction with the page, including login flows and content that appears only after the page runs.

web-browser — Skill by aretw0

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 skills/aretw0/agents-lab/web-browser

Any agent

npx skills add aretw0/agents-lab --skill web-browser

Clone the repo

git clone --depth 1 https://github.com/aretw0/agents-lab

Made for: Claude Code, Codex.

Per session 36 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 838 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 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.

Model	Per session	Once invoked
Fable 5	$0.00036	$0.00838
Opus 5	$0.00018	$0.00419
Sonnet 5	$0.00007	$0.00168
Haiku 4.5	$0.00004	$0.00084

Measured 2d ago against content hash dfc0fbc0f93a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

web-browser scanned grade A 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.

The scan reads SKILL.md. This mod also ships 10 executable files (scripts/cdp.js, scripts/dismiss-cookies.js, scripts/eval.js, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- Do **not** start from `curl`/`wget`/`python requests`/`r.jina.ai` or generic shell scraping.

packages/web-skills/skills/web-browser/SKILL.md · 111 lines

How it starts

The opening of the file, as written. The whole thing — 111 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Web Browser Skill

Browser automation via Chrome DevTools Protocol (CDP). All scripts live in ./scripts/ relative to this skill.

When to Prefer This Skill

Use web-browser for interactive web tasks:

open/navigate pages and tabs
click buttons/links
fill forms and login flows
inspect dynamic UI state rendered in-browser

For non-interactive tasks (searching links, extracting page text, fetching docs), prefer web_search / fetch_content instead.

Scoped Hard Trigger Policy

Enforce browser-first (hard-by-scope) when BOTH conditions are present:

Prompt has interactive intent (open, navigate, click, fill, login, submit, tab, form, button), and
Target is a sensitive/challenge-prone domain (e.g. npmjs.com) or the user reports Cloudflare/bot blocking.

When this trigger fires:

Start with CDP flow (start.js → nav.js → interaction/eval scripts).
Do not start from curl/wget/python requests/r.jina.ai or generic shell scraping.
Only fallback after explicit CDP failure evidence (at least 2 failed attempts), and report the fallback reason.

Dependency: @aretw0/web-skills declares ws at the package root. Published installs should not require per-skill setup. When running from this repository source, run pnpm install at the repository root before using the scripts.

Start Chrome

node ./scripts/start.js              # Fresh profile
node ./scripts/start.js --profile    # Copy your real profile (cookies, logins)

Starts Chrome on :9222 with remote debugging enabled.

If Chrome is in a non-standard location:

BROWSER_BIN=/path/to/chrome node ./scripts/start.js

Navigate

node ./scripts/nav.js https://example.com           # Current tab
node ./scripts/nav.js https://example.com --new     # New tab

Evaluate JavaScript

node ./scripts/eval.js 'document.title'
node ./scripts/eval.js 'document.querySelectorAll("a").length'
node ./scripts/eval.js 'JSON.stringify(
  Array.from(document.querySelectorAll("a"))
    .map(a => ({ text: a.textContent.trim(), href: a.href }))
    .filter(l => l.href.startsWith("https://"))
)'

Read the full file on GitHub · 111 lines

Files

What ships with it

10 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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.

2d ago First seen · 111 lines · 36 tokens per session scan A dfc0fbc0f93a

Subscribe to this mod's changes

web-browser is a skill published in the GitHub repository aretw0/agents-lab (11 stars, last pushed 2mo ago), licensed MIT. It adds 36 tokens to every session and 838 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Browser automation LLM agent-tools ai-agents automation coding-agent developer-tools

gpt-control

Use when the user asks to get an independent ChatGPT web review, continue a GPT-Control conversation, inspect a model run, or generate or iterate on an image through a configured browser driver. Not for facts the working tree or a local test can answer directly.

wolfiesch/gpt-control · 56 tokens

cua-driver

Drive a native GUI app (macOS, Windows, Linux) via the Qwen Cua Driver CLI (default) or MCP server; snapshot its accessibility tree, act through snapshot-bound element tokens, native menu paths, exact window geometry, or pixel coordinates, and verify from fresh state. Use when the user asks you to operate, drive…

QwenLM/qwen-code · 88 tokens

computer-use

Control local desktop applications through Computer Use for tasks that require reading or operating app UI. Prefer purpose-built connectors, APIs, or CLIs when available.

QwenLM/qwen-code · 33 tokens

webapp-testing

Start/reuse a local app, wait for readiness, inspect rendered state/console/network, act from observed selectors, and verify with evidence.

Hmbown/CodeWhale · 32 tokens

sharing-results

The user wants to share or hand off what was found in a video — "send this analysis to my team", "make a page I can share", "export the findings", "give me something I can attach to the ticket". Use this to render a self-contained offline HTML viewer page with the frames, transcript, and every cached answer with its…

oxbshw/watch-skill · 75 tokens

webbrowser

Use the native webbrowser module for web search, browsing, authenticated website interaction, forms, file uploads and downloads, screenshots, and user handoff.

aduermael/herm · 33 tokens

web-browser