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 agentmods add skills/leeguooooo/chrome-use/corenpx skills add leeguooooo/chrome-use --skill coregit clone --depth 1 https://github.com/leeguooooo/chrome-useWhat 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 | $0.00112 | $0.16995 |
| Opus 5 | $0.00056 | $0.08497 |
| Sonnet 5 | $0.00022 | $0.03399 |
| Haiku 4.5 | $0.00011 | $0.01699 |
Grade A, and why
core 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 yesterday.
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.
| Specific facts from a static or public page | `WebFetch` or `curl` (no browser) | How it starts
The opening of the file, as written. The whole thing — 1,343 lines — stays where its author put it; the contents beside it link to each section on GitHub.
chrome-use core
Fast browser automation CLI for AI agents. Chrome/Chromium via CDP, no
Playwright or Puppeteer dependency. Accessibility-tree snapshots with compact
@eN refs let agents interact with pages in ~200-400 tokens instead of
parsing raw HTML.
Most normal web tasks (navigate, read, click, fill, extract, screenshot) are covered here. Load a specialized skill when the task falls outside browser web pages — see When to load another skill.
Hit a rough edge? Please report it. If a command surprised you — a confusing error, a stale
@ref, an occluded click, a flaky wait, a missing feature, or anything that cost you extra turns — open a quick issue at https://github.com/leeguooooo/chrome-use/issues with the exact command and what happened vs. what you expected. Agent-filed friction reports are how this tool gets sharper; a 30-second issue is genuinely valuable.(Failures are also auto-logged locally — run
chrome-use frictionto see what's been painful, by command/category/host. Local only, never uploaded; opt out withAGENT_BROWSER_NO_FRICTION_LOG=1.)
The core loop
chrome-use open <url> # 1. Open a page
chrome-use snapshot -i # 2. See what's on it (interactive elements only)
chrome-use click @e3 # 3. Act on refs from the snapshot
chrome-use snapshot -i # 4. Re-snapshot after any page change
Refs (@e1, @e2, ...) are stable for the same backend DOM node across
successive snapshots of one document, so inserting or removing a modal no
longer renumbers every later control. Navigation and tab switches hard-reset
the identity map. Re-snapshot after those boundaries, and whenever you need to
discover newly rendered controls.
@refs self-heal across re-renders — you don't need to re-snapshot for every minor DOM churn. Each ref records a fingerprint (role + accessible name + ancestor path); if its node is gone when you use it, chrome-use automatically relocates to the matching element on the current page and proceeds. So after a React/Vue list re-render that keeps the same labels,
click @e3still hits the right element. If the element is genuinely gone, it refuses (loud error) rather than click the wrong node — it never silently mis-targets. Re-snapshot when you navigated, switched tabs, or need refs for newly created elements.
What ships with it
11 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.
- references/authentication.md 8.9 KB
- references/commands.md 30 KB
- references/profiling.md 3.3 KB
- references/proxy-support.md 4.8 KB
- references/session-management.md 4.5 KB
- references/snapshot-refs.md 5.7 KB
- references/trust-boundaries.md 4.0 KB
- references/video-recording.md 3.7 KB
- templates/authenticated-session.sh 3.5 KB runs code
- templates/capture-workflow.sh 1.7 KB runs code
- templates/form-automation.sh 1.7 KB runs code
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.
- yesterday First seen · 1,343 lines · 112 tokens per session scan A 1367e92769c7
core is a skill published in the GitHub repository leeguooooo/chrome-use (156 stars, last pushed 2d ago), licensed Apache-2.0. It adds 112 tokens to every session and 16,995 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). It comes from a forked repository.
Other skills, from other repositories
ultimate-playwright
Browser automation via ultimate-playwright-mcp. Use when the agent needs to control a shared Chrome browser — open tabs, navigate, click, type, fill forms, take snapshots. Handles multi-user tab group isolation so multiple sessions can share one browser without interference. Use for web browsing, scraping, form…
surf
Control Chrome browser via CLI for testing, automation, and debugging. Use when the user needs browser automation, screenshots, form filling, page inspection, network/CPU emulation, DevTools streaming, or AI queries via ChatGPT/Gemini/Perplexity/Grok/AI Studio.
surf-codebase
Navigate and modify surf-cli codebase - Chrome extension + native host for AI browser automation. Use for surf-cli code work, architecture questions, implementing browser control/CDP/accessibility/network features.
Agent Browser Automation
Fast Rust-based headless browser automation CLI with Node.js fallback for AI agents, featuring navigation, clicking, typing, snapshots, and structured commands optimized for agent workflows.
real-browser-control
Connect an AI coding agent to the user's REAL Chrome via Real Browser MCP (local MCP server + Chrome extension over localhost WebSocket). Use for live UI verification, already-open tabs, existing cookies/SSO/2FA, staging admin panels, and any task where headless Playwright or a cloud agentic browser would miss the…
aginxbrowser
Browser engine for AI agents: fetch JS-rendered and Cloudflare-protected pages as clean markdown, run 5-engine aggregated web search (Baidu, Bing, Sogou, WeChat, Google), take screenshots as visual input, extract structured data from SPAs, and drive multi-step interactions (click, type, fill forms, login, paginate)…