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 metahub-tech/agent-fleet --skill using-human-browsergit clone --depth 1 https://github.com/metahub-tech/agent-fleetWrote 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/metahub-tech/agent-fleet/using-human-browser)<a href="https://agentmods.dev/skills/metahub-tech/agent-fleet/using-human-browser"><img src="https://agentmods.dev/badge/skills/metahub-tech/agent-fleet/using-human-browser/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/metahub-tech/agent-fleet/using-human-browser"><img src="https://agentmods.dev/badge/skills/metahub-tech/agent-fleet/using-human-browser.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 | $0.00090 | $0.01681 |
| Opus 5.5 | $0.00036 | $0.00672 |
| Sonnet 5.5 | $0.00018 | $0.00336 |
| Haiku 4.5 | $0.00009 | $0.00168 |
Grade A, and why
using-human-browser 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 1mo 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.
How it starts
The opening of the file, as written. The whole thing — 47 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Using human_browser
human_browser is a self-built capability: it launches the host's real daily Google Chrome with no debug port and no automation flags, so there are zero automation traces — navigator.webdriver stays false, there is no CDP interface to probe, and input is genuine OS-level input. You then operate the page the way a person does: look at a screenshot, click coordinates, type.
This is the moat: it reuses the device's OS-level control (core tools), not a browser-automation framework.
human_browser vs agent_browser (routing)
- human_browser (this skill): real Chrome, real logged-in identity, zero automation traces. Page localization:
human_dom(read-only DOM locator companion — exact coordinates, see using-human-dom) or screenshot+coordinates. Use when acting as the human on real accounts / identity. - agent_browser: Playwright-driven, DOM snapshot+ref, fast — but has automation traces and uses an isolated profile (NOT the human's identity, NOT the user's login). Use for testing / scraping / browsing-to-learn on throwaway isolated profiles.
★ Rule for real accounts (read this — it's the #1 footgun): a real account / identity → human_browser + human_dom, from the FIRST navigation (including the login scan) through every op. NEVER touch agent_browser for that account — not even to "just browse and check" first. Why: agent_browser logs into its isolated profile (~/.fleet/agent-browser-profile), invisible to human_browser; if you scan/login there, your later human_browser runs see a logged-OUT browser → you re-login forever (this was a real bug: a publisher re-scanned every hour for 12h). human_dom now gives human_browser DOM-level precision, so there is no reason to use agent_browser on a real account.
登录态跨 run 持久 + resume（recurring operator 必读）
A recurring operator (an agent driving a real account every run / cron tick) must reuse ONE login across runs — the user scans once, not every run:
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.
- 1mo ago First seen · 47 lines · 90 tokens per session scan A c05593159d82
using-human-browser is a skill published in the GitHub repository metahub-tech/agent-fleet (2 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 90 tokens to every session and 1,681 once invoked, about $0.0004 per session on Opus 5.5. 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
web-exfiltration-detection
Detect data exfiltration via URL path encoding and chained webfetch navigation. Covers fake trusted UI injection, letter-level URL path exfiltration, and multi-hop navigation hijacking. Use when the agent has web/URL fetch capability and stores user memory or personal context.
browserwing-executor
Control browser automation through HTTP API. Supports page navigation, element interaction (click, type, select), data extraction, accessibility snapshot analysis, screenshot, JavaScript execution, and batch operations.
dev-browser
Browser automation with persistent page state. Use when users ask to navigate websites, fill forms, take screenshots, extract web data, test web apps, or automate browser workflows. Trigger phrases include "go to [url]", "click on", "fill out the form", "take a screenshot", "scrape", "automate", "test the website"…
web-search
This skill should be used when users need to search the web for information, find current content, look up news articles, search for images, or find videos. It uses DuckDuckGo's search API to return results in clean, formatted output (text, markdown, or JSON). Use for research, fact-checking, finding recent…
langbot-testing
Test LangBot WebUI and core product flows with an automated browser and backend logs. Use when validating the configured LangBot frontend, pipeline Debug Chat, model provider setup and test buttons, bot and knowledge-base UI flows, or troubleshooting failed LangBot end-to-end tests.
browser-automation
Drive the PenguinHarness agent browser — the desktop app's built-in browser or the user's own Chrome — from the shell with penguin browser: open pages, read them as simplified HTML or text, act with JavaScript and trusted clicks and typing, and pull structured data out of them (orders, search results, tables), signed…