← Back to Skills Marketplace
webcli
by
Erdinç Cürebal
· GitHub ↗
· v0.2.1
767
Downloads
0
Stars
0
Active Installs
4
Versions
Install in OpenClaw
/install webcli
Description
Browse the web, read page content, click buttons, fill forms, take screenshots, and get accessibility snapshots using the webcli headless browser. Use when t...
Usage Guidance
This skill appears internally consistent for headless browser automation. Before installing or running it: (1) review and trust the npm package and its GitHub repo (supply‑chain risk); (2) run the tool in a sandbox or isolated environment if possible; (3) avoid directing the agent to log into or visit pages containing highly sensitive data unless you trust the agent and skill, because cookies/localStorage/session exports could leak credentials; (4) be aware the prerequisites will download browser binaries (Playwright) and install a global npm package — only proceed if you trust those sources.
Capability Analysis
Type: OpenClaw Skill
Name: webcli
Version: 0.2.1
The skill bundle is classified as suspicious because it grants the AI agent access to a powerful web automation tool (`webcli`) with capabilities that, while legitimate for the tool's purpose, pose significant security risks if misused. Specifically, commands like `webcli eval` (arbitrary JavaScript execution), `webcli upload` (uploading local files), `webcli cookie export`, `webcli storage get`, and `webcli state save` (accessing and saving sensitive browser data) could be leveraged through prompt injection against the agent to perform data exfiltration or unauthorized actions. The `SKILL.md` file itself does not contain malicious instructions but documents these risky capabilities.
Capability Assessment
Purpose & Capability
The name/description (web browsing, clicking, filling forms, screenshots, accessibility snapshots) align with the SKILL.md commands and examples. The prerequisite npm package and Playwright browser install are expected for a CLI headless browser.
Instruction Scope
Instructions focus on web navigation and interaction, which is in‑scope. They also include state/cookie export, localStorage access, file upload, and saving/restoring session.json — these are required for realistic browser automation but can expose site sessions or user files if misused.
Install Mechanism
There is no registry install spec; the SKILL.md tells users to run 'npm install -g @erdinccurebal/webcli' and 'npx playwright install chromium'. Using an npm-scoped package and Playwright is normal for this tool, but it does mean executing third‑party code and downloading browser binaries from the network (supply-chain risk).
Credentials
The skill declares no required environment variables or config paths. However, its functionality (cookies, localStorage, state export, file upload) can capture or persist sensitive data from sites visited; this is functionally necessary but worth noting from a privacy/credential exposure perspective.
Persistence & Privilege
The skill does not request always:true, system config changes, or other skills' credentials. It is instruction-only and does not demand persistent platform privileges. Autonomous invocation is enabled by default (normal), so consider agent policies that control which URLs or actions the agent may perform.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install webcli - After installation, invoke the skill by name or use
/webcli - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.2.1
Update website URL to webcli.erdinc.curebal.dev
v0.2.0
Add 28 new commands: snapshot with refs, dblclick, hover, check, uncheck, drag, upload, scroll, title, url, value, count, box, styles, visible, enabled, checked, pdf, console, errors, dialog, frame, storage, device, state save/load. Config extensions for geolocation, colorScheme, offline, extraHeaders.
v0.1.1
- Reduced and simplified the skill description for clarity.
- Added official homepage and GitHub repository links.
- Removed references to network logging, cookies, and JavaScript evaluation commands from the documentation.
- Streamlined the commands section to focus on core navigation, interaction, and tab management features.
- Omitted advanced troubleshooting, headed mode, and page debugging guidance.
- Retained essential best practices and usage notes.
v0.1.0
webcli 0.1.0 — Initial Release
- Introduces headless browser automation via the webcli command-line tool.
- Supports navigation, content reading, scraping, form filling, clicking, taking screenshots, and managing tabs/sessions.
- Provides commands for cookies, network monitoring, and browser configuration.
- Includes recommendations, troubleshooting, and best practices for robust web automation tasks.
- Requires installation of webcli and Playwright, with headed mode available for bot-protected sites.
Metadata
Frequently Asked Questions
What is webcli?
Browse the web, read page content, click buttons, fill forms, take screenshots, and get accessibility snapshots using the webcli headless browser. Use when t... It is an AI Agent Skill for Claude Code / OpenClaw, with 767 downloads so far.
How do I install webcli?
Run "/install webcli" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is webcli free?
Yes, webcli is completely free (open-source). You can download, install and use it at no cost.
Which platforms does webcli support?
webcli is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created webcli?
It is built and maintained by Erdinç Cürebal (@erdinccurebal); the current version is v0.2.1.
More Skills