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.
npx skills add HanzMaulwurf/tex-browse-use --skill browse-usegit clone --depth 1 https://github.com/HanzMaulwurf/tex-browse-useWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/hanzmaulwurf/tex-browse-use/browse-use)<a href="https://agentmods.dev/skills/hanzmaulwurf/tex-browse-use/browse-use"><img src="https://agentmods.dev/badge/skills/hanzmaulwurf/tex-browse-use/browse-use/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/hanzmaulwurf/tex-browse-use/browse-use"><img src="https://agentmods.dev/badge/skills/hanzmaulwurf/tex-browse-use/browse-use.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 5.1 | $0.00082 | $0.00566 |
| Opus 5 | $0.00041 | $0.00283 |
| Sonnet 5 | $0.00016 | $0.00113 |
| Haiku 4.5 | $0.00008 | $0.00057 |
Grade A, and why
browse-use scanned grade A with 0 findings 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 11d 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Browse-use (TEX)
Operate any web app by driving a real browser when there is no API, MCP, CLI, or A2A path. TEX runs a fallback ladder — cheap DOM agent first, escalating to vision + stealth — and reports back what it found or did.
When to use this
Reach for browse_use when the task genuinely needs a browser:
- Logging into a web app and reading/changing something there
- Extracting data from a dashboard or report that has no export/API
- Filling in or submitting a web form
- Clicking through a multi-step portal flow
- Verifying what a live page actually displays
Do not use it when a normal API, WebFetch, or CLI already does the job — those are faster and cheaper. Browse-use is the last resort for UI-only apps.
How to use it
The plugin exposes two MCP tools (server tex-browse-use):
browse_use(task, url?, app?)— give a clear, specific task in plain language and (usually) a starturl. It blocks until done and returns the agent's findings plus a run summary (status, gear/strategy, steps, tokens, final URL). Optionalappreuses a saved login session, learned skills, and stored credentials for that app (e.g.app: "acme-crm").tex_health()— check the engine is up and which capabilities (vision, DOM, stealth, compiled replay) are available.
Write the task so the agent knows both what to do and what to report back, e.g.
"Open the orders page, find the most recent order, and report its order number and total."
Prerequisite: the engine must be running
browse_use talks to a local TEX engine over HTTP. If it returns an "engine not reachable" error, start the engine first:
scripts/tex-up.sh
Then retry. Run tex_health to confirm it's up. See the plugin README for one-time setup (.env with an LLM provider key, npm install, playwright install chromium).
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.
- 11d ago First seen · 42 lines · 82 tokens per session scan A e0d539ba775b
browse-use is a skill published in the GitHub repository HanzMaulwurf/tex-browse-use (0 stars, last pushed 2mo ago), licensed MIT. It adds 82 tokens to every session and 566 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
playwright-cli
Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
webapp-testing
Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.
screen-use
A Windows desktop automation skill for clicking controls, entering text, using keyboard shortcuts, scrolling, and taking screenshots. It can also carry out multi-step tasks through visual guidance.
x402
Set up Browser Use Cloud payments with x402 — pay per request from a crypto wallet (USDC on Base mainnet), no signup or API key. Two setups it works out up front — "just use it" (set up a wallet so you or Claude Code can run cloud browser tasks paid from the wallet — Claude writes and runs throwaway scripts, nothing…
browser-use
Direct browser control via CDP for web interaction: automation, scraping, testing, screenshots, and site/app work.
blocked-page-recovery
Use when a fetch fails: 403/429, paywall, WAF, bot wall.