← Back to Skills Marketplace
tangjing

Token Efficient Web Operations en

by Jing Tang · GitHub ↗ · v1.0.1 · MIT-0
cross-platform ⚠ suspicious
225
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install page-js-operation-en
Description
Operate web page elements via CDP and Page JS Extension, save 95% Token (for UI automation testing ONLY, NOT a web browsing tool)
Usage Guidance
This skill appears coherent for automated UI testing, but take these precautions before use: 1) Manually review the linked Page JS extension source code (https://github.com/TangJing/openclaw_access_web_page_js) before installing; 2) Only run the skill in controlled CI/test environments, never on banking, payment, healthcare, login or other sensitive pages; 3) Ensure your agent/LLM setup does not send raw DOM content or secrets to external services — prefer local LLMs or block outbound LLM calls during tests; 4) Audit the extension's sensitive-field filtering (covers non-INPUT or custom components?) and test edge cases; 5) If you need stronger guarantees, disable autonomous invocation for this skill or restrict it to manually triggered runs.
Capability Analysis
Type: OpenClaw Skill Name: page-js-operation-en Version: 1.0.1 The skill enables browser automation via the Chrome DevTools Protocol (CDP) and requires the manual installation of a third-party Chrome extension from GitHub (https://github.com/TangJing/openclaw_access_web_page_js). It utilizes high-privilege `Runtime.evaluate` calls to execute arbitrary JavaScript on web pages, which is a significant security risk. Although the documentation in SKILL.md includes extensive privacy warnings and claims to skip sensitive fields like passwords during data collection, the reliance on external, unverified code and the potential for data exfiltration to LLM services warrant a suspicious classification.
Capability Assessment
Purpose & Capability
Name/description, declared dependencies (Page JS Extension + CDP), and the SKILL.md instructions all focus on browser automation and element operations; there are no unrelated env vars, binaries, or installs requested.
Instruction Scope
Instructions are consistent with UI automation (exportData, searchElementsByKey, click, fill). The doc explicitly forbids sensitive sites and shows code to skip password/hidden/file inputs, which is appropriate. However, the agent-level guidance admits element keys/data may be sent to external LLMs — that creates a real exfiltration risk if the agent is misconfigured or the LLM call is not local. Also review whether the 'skip sensitive fields' logic covers all edge cases (non-INPUT elements, custom controls, attributes).
Install Mechanism
This is instruction-only (no install spec). The only install step is a manual extension install from a GitHub repo; that is reasonable for this purpose but requires the user to inspect the extension source before trusting it.
Credentials
No environment variables, secrets, or system paths are requested. The skill does not ask for unrelated credentials or persistent tokens.
Persistence & Privilege
always:false (normal). Model invocation is allowed (default), so the agent could autonomously run CDP commands that result in element data being sent to an LLM — this is expected for automated skills but increases blast radius; limit autonomous use in sensitive contexts.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install page-js-operation-en
  3. After installation, invoke the skill by name or use /page-js-operation-en
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.1
- Clarified this Skill is exclusively for automation testing, not general web browsing or personal use. - Strengthened privacy field policy: password, hidden, and file fields are never collected or read; write access only by explicit user command. - Updated all usage recommendations and restrictions to emphasize the Skill is NOT for handling credentials or personal data. - Improved technical explanations and documentation structure to highlight the security model and explicit user responsibility. - No code or functional changes; documentation update only.
v1.0.0
Initial release of the Page JS CDP Web Operation Skill - Enables UI automation testing by controlling web pages via Chrome DevTools Protocol and a custom Chrome extension. - Significantly reduces token usage for page operations (up to 95% less). - Explicitly designed for testing and demo environments; not intended for production or sensitive data pages. - Includes strict privacy and security guidelines—prohibits use on banking, login, or any personal data pages. - Offers easy integration and clear API templates for common actions (click, fill input, select dropdown, etc.). - Requires manual installation of the accompanying Chrome extension and CDP-compatible environment.
Metadata
Slug page-js-operation-en
Version 1.0.1
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 2
Frequently Asked Questions

What is Token Efficient Web Operations en?

Operate web page elements via CDP and Page JS Extension, save 95% Token (for UI automation testing ONLY, NOT a web browsing tool). It is an AI Agent Skill for Claude Code / OpenClaw, with 225 downloads so far.

How do I install Token Efficient Web Operations en?

Run "/install page-js-operation-en" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Token Efficient Web Operations en free?

Yes, Token Efficient Web Operations en is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Token Efficient Web Operations en support?

Token Efficient Web Operations en is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Token Efficient Web Operations en?

It is built and maintained by Jing Tang (@tangjing); the current version is v1.0.1.

💬 Comments