Automate repeatable browser-based workflows such as navigating dashboards, filling forms, and exporting data using deterministic execution paths.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install browser-task-runner
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install browser-task-runner using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Browser Task Runner?
The Browser Task Runner skill provides AI coding agents with a structured methodology to plan and execute reliable, multi-step browser-based workflows. Instead of relying on fragile, improvised interactions, this skill emphasizes deterministic execution paths on the local machine to handle routine operations like opening URLs, navigating complex dashboards, logging into internal tools, and capturing state snapshots.
By treating web environments as naturally volatile, the skill enables Openclaw Skills users to gracefully handle public read-only pages, leverage advanced local automation frameworks, or fall back to native macOS UI scripting. It mitigates automation risks by isolating destructive actions and mapping precise hybrid boundaries where manual user takeover is required.
Browser Task Runner Use Cases
- Automating Routine Authentications: Logging into internal dashboards or operational tools that follow fixed navigation pathways.
- Data Extraction & Generation: Navigating to report pages to export structured CSV or PDF data, or scraping visible page summaries.
- State Documentation: Capturing high-fidelity screenshots of specific page states for testing, compliance, or logging.
- Repetitive Form Submission: Automatically populating and submitting multi-field web forms with precision.
How Browser Task Runner Works
- Outcome Confirmation: The agent confirms the target website URL and verifies the exact desired outcome before initiating any browser instances.
- Path Identification: Evaluates available execution methods, prioritizing direct web fetches for public data, local browser automation tooling next, and macOS AppleScript/UI scripting as a fallback.
- Security & Gate Check: Scans the workflow requirements for authentication layers, Multi-Factor Authentication (MFA), CAPTCHAs, or destructive account-changing actions.
- Execution Safeguard: Prompts the user for explicit confirmation if the workflow involves any risk of permanent state modifications.
- Micro-Flow Execution: Executes the sequence in the smallest viable increments, isolating failure points and indicating precisely where human assistance is required.
Browser Task Runner Setup
To deploy this capability within your local agent workspace, register the schema definition in your project configuration. Ensure your local machine has the necessary automation dependencies installed.
# Initialize and register the browser task runner skill
openclaw skills install browser-task-runner
# Verify that local browser dependencies (Playwright, Puppeteer, or AppleScript permissions) are configured
openclaw skills verify browser-task-runner
Browser Task Runner Data Schema & Taxonomy
The skill manages workflows by organizing configurations into deterministic automation blocks and reporting outcomes via a standardized schema.
| Block Element | Type | Description |
|---|---|---|
Goal |
String | Clear objective or target URL of the automated browser task. |
Automation Path |
String | Selected framework or execution tool (e.g., direct web tool, Playwright, macOS UI Scripting). |
What Succeeded |
List | Granular breakdown of individual steps executed successfully. |
Where It May Break |
List | Highlighted fragile steps, popups, or anti-bot/MFA checkpoints encountered. |
Next Improvement |
String | Recommendations for shifting logic to backend scripts or hardening the workflow. |
Browser Task Runner Advanced Features
- Hybrid Human-in-the-Loop Handouts: Automatically pauses and yields control to the user when encountering volatile components like CAPTCHAs, MFA prompts, or payment screens.
- Multi-tiered Fallback Engine: Dynamically shifts down from robust browser APIs to native OS-level UI scripting when dedicated automation binaries are absent.
- Destructive Action Guards: Built-in protocol to intercept and mandate manual authorization before committing changes to account configurations or stateful profiles.
SKILL.md
Loading
Related Openclaw Skills
GPT Image Generation
An advanced Openclaw skill that automates ChatGPT GPT-Image-2 via Chrome DevTools Protocol to deliver photorealistic images, precise text rendering, and complex composition.
Daily Learning Framework
A unified daily learning and knowledge digestion framework that empowers AI agents to autonomously research topics, build structured markdown notes, and sync knowledge bases.
The Grand Design
An executable knowledge toolkit based on Stephen Hawking and Leonard Mlodinow's The Grand Design to explore the fundamental laws of physics, quantum mechanics, and cosmology.
The Miracle of Mindfulness AI Skill
An executable AI coding agent skill that delivers structured mindfulness teachings and practical meditation frameworks inspired by Thich Nhat Hanh.
Mac Maintenance
A terminal-first macOS inspection and optimization tool built for non-destructive system maintenance and diagnostic reporting.
Smart Search
Smart Search provides AI agents with a robust multi-engine web search skill featuring automatic language-based routing, seamless fallbacks, and markdown extraction capabilities.
Featured*
