browser-work

Explore target websites and generate a pitfall-prevention guide for browser automation.

Updated Apr 14, 2026
One-click install
npx skills add https://github.com/sr-ai-dev/sr-harness --skill browser-work
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: browser-work
Source: https://github.com/sr-ai-dev/sr-harness/tree/main/skills/browser-work
Command: npx skills add https://github.com/sr-ai-dev/sr-harness --skill browser-work

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill solves the problem of unreliable browser automation by reducing trial-and-error on unfamiliar sites through recon-first exploration and a pitfall-prevention guide before delegation.

Core Features & Use Cases

  • Recon-first orchestration: Uses chromux to navigate, snapshot, and walk the full workflow once to discover stable element references and site behaviors.
  • Guide generation for reliable execution: Writes a session-specific guide file (WORK_DIR/guide.md) capturing steps, expected @ref patterns, pitfalls, and bot-detection cautions.
  • Delegated execution to a purpose-built agent: Launches the browser-explorer subagent with the recon guide embedded, instructing it to re-snapshot after changes and stop on CAPTCHA/rate limits.
  • Use cases: Completing multi-step tasks on websites where layouts, element references, or dynamic content make direct automation brittle (e.g., forms, multi-page workflows, LinkedIn-style interactions).

Quick Start

Ask the AI to use the browser-work flow to automate your site task by specifying the target site and what you want done.

Frequently Asked Questions about browser-work

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate multi-step website workflows without breaking on dynamic content?

Recon-first browser automation solves this by exploring the target site once to document stable element references and generate a pitfall-prevention guide before delegating execution. This ensures reliable @ref targeting for dynamic multi-page interactions.

Why does my browser automation fail on unfamiliar sites with brittle element selectors?

Browser automation fails on unfamiliar sites due to trial-and-error targeting. Recon-first exploration walks the full workflow, snapshots site behaviors, and writes a session-specific guide capturing stable @ref patterns and pitfalls to prevent brittle execution.

How do I handle bot detection and CAPTCHAs during automated web interactions?

Automated web interactions handle bot detection by applying wait/delay strategies and explicit handling for CAPTCHA or rate-limit failures. The delegated browser-explorer agent stops execution upon encountering these blocking constraints.

Do I need chromux to run recon-first web orchestration tasks?

Yes, chromux availability with headless launch is required for recon-first web orchestration. It drives the initial site navigation, snapshotting, and workflow exploration to discover stable element references before delegating the actual execution.

Can I automate complex form completion on sites with dynamic layouts?

Yes, complex form completion on dynamic layouts is supported through snapshot-driven interaction rules. The session-scoped guide captures expected @ref patterns and pitfalls, enabling the browser-explorer agent to re-snapshot after changes and execute reliably.

What is the best way to complete multi-page workflows on sites like LinkedIn using browser automation?

The best way to complete multi-page workflows on sites like LinkedIn is recon-first automation. It explores the target site, generates a session-specific guide with expected @ref patterns and bot-detection cautions, then delegates execution to re-snapshot after changes.