← Back to Skills Marketplace
My Agent Browser
by
liujiang817
· GitHub ↗
· v1.0.0
· MIT-0
251
Downloads
0
Stars
2
Active Installs
1
Versions
Install in OpenClaw
/install my-agent-browser
Description
A fast Rust-based headless browser automation CLI with Node.js fallback that enables AI agents to navigate, click, type, and snapshot pages via structured co...
Usage Guidance
This SKILL.md documents a CLI wrapper that you would normally install via npm. Before installing or running it: 1) verify the upstream package and publisher on the npm registry (ensure the package name and publisher match the github repo referenced in SKILL.md), 2) confirm the repository and package integrity (check the GitHub repo, release tags, and maintainer identity), 3) avoid blindly running global npm installs on sensitive machines — prefer a container, VM, or non-root user, 4) note that the skill bundle's registry metadata (ownerId/slug) does not match the included _meta.json, and there is no homepage recorded in the registry entry — this provenance mismatch is the primary reason for caution, and 5) if you need higher assurance, ask the publisher to provide a verifiable source URL or an official npm package name before installing.
Capability Analysis
Type: OpenClaw Skill
Name: my-agent-browser
Version: 1.0.0
The skill bundle provides a comprehensive and well-documented interface for the 'agent-browser' CLI, a legitimate tool for headless browser automation. It includes standard commands for web navigation, interaction, and session management (such as cookie handling and state saving). While the tool inherently possesses high-risk capabilities like JavaScript execution and session data access, these are strictly aligned with its stated purpose of web automation and lack any indicators of malicious intent, unauthorized data exfiltration, or prompt injection in SKILL.md.
Capability Assessment
Purpose & Capability
The SKILL.md describes an agent-browser CLI and the required binaries (node, npm) and npm-based install match that purpose. However, the description claims a 'Rust-based' CLI with Node fallback while the install/usage is npm-centric; additionally registry metadata (skill ownerId/slug) does not match the _meta.json slug/owner inside the bundle and there is no declared homepage or source in the registry entry. These provenance inconsistencies are not fatal but are unexpected and worth verifying.
Instruction Scope
Runtime instructions are narrowly scoped to installing and running the agent-browser CLI and its commands (navigate, snapshot, click, screenshot, record, etc.). The SKILL.md does not instruct the agent to read unrelated files, exfiltrate data to third-party endpoints, or access credentials. Commands assume a locally installed CLI and act on pages, which is consistent with the stated purpose.
Install Mechanism
There is no install spec in the registry (instruction-only), and the SKILL.md recommends npm -g install and a git clone of a GitHub repo. Using npm -g writes system-global packages and may require elevated permissions; the SKILL.md points to github.com/vercel-labs/agent-browser (a plausible upstream) but the registry entry lacks a homepage/source and internal _meta.json differs from registry metadata. This mismatch increases provenance risk when following global npm install instructions.
Credentials
The skill requires node and npm but declares no environment variables, secrets, or config paths. The lack of requested credentials is proportional to the stated browser-automation purpose.
Persistence & Privilege
The skill is not always-enabled and allows user invocation; it does not request elevated platform privileges or persistently modify other skills. This is appropriate for an instruction-only CLI wrapper.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install my-agent-browser - After installation, invoke the skill by name or use
/my-agent-browser - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
- Initial release of Agent Browser, a fast Rust-based headless browser automation CLI with Node.js fallback.
- Enables AI agents to automate navigation, clicks, typing, snapshots, form filling, and data extraction on web pages via structured Bash commands.
- Supports advanced actions like taking screenshots, recording video demos, tab/window management, handling cookies and storage, and network request interception.
- Provides semantic element locators by role, text, or label, in addition to using element references from interactive snapshots.
- Includes state management for persistent sessions and authentication.
- Extensive command set for mouse and keyboard control, JavaScript evaluation, and accessibility/tree inspections.
Metadata
Frequently Asked Questions
What is My Agent Browser?
A fast Rust-based headless browser automation CLI with Node.js fallback that enables AI agents to navigate, click, type, and snapshot pages via structured co... It is an AI Agent Skill for Claude Code / OpenClaw, with 251 downloads so far.
How do I install My Agent Browser?
Run "/install my-agent-browser" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is My Agent Browser free?
Yes, My Agent Browser is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does My Agent Browser support?
My Agent Browser is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created My Agent Browser?
It is built and maintained by liujiang817 (@liujiang817); the current version is v1.0.0.
More Skills