# Forge Reusable Website Automation Skills

Repeated website exploration wastes time and produces fragile automation. This skill inspects browser behavior, verifies extraction or operation paths, and generates reusable Skill packages.

## Install

```bash
npx skillstore add browser-act/browser-act-skill-forge
```

## Metadata

- Status: approved
- Slug: browser-act-browser-act-skill-forge
- Version: 1.0.6
- Author version: 1.0.6
- Skillstore revision: r3
- Version status: valid
- Tree hash: 4479e3cacf88342bbe619a626af73f62c47e0a40b8308940a0805405932e1555
- Author: browser-act
- GitHub username: browser-act
- License: MIT
- Repository: https://github.com/browser-act/skills/tree/main/browser-act-skill-forge
- Ref: a39a91716eadede5f4cdefd78178fed4e837a128
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: confirmation\_required
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: network, external\_commands
- Quality score: 38
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/browser-act-browser-act-skill-forge
- Manifest: https://skillstore.pages.dev/api/skills/browser-act-browser-act-skill-forge/manifest

## Capabilities

- Decomposes a website request into reusable extraction or operation capabilities and confirms the execution plan.
- Discovers target-site APIs through browser traffic, then falls back to network capture, DOM automation, or visual workflows.
- Verifies pagination, parameters, selectors, and enumeration sources against the active website.
- Packages verified browser JavaScript into Python wrappers and generates capability-focused SKILL.md files.
- Tests generated skills, installs them when possible, and can execute the original batch task.

## Use Cases

- Build Repeatable Data Collection: Create a verified extraction skill for recurring lists, details, filters, and pagination.
- Automate Browser Operations: Package a validated form or account workflow into reusable browser steps.
- Publish Site-Specific Skills: Convert one browser exploration into documented, tested Skill packages with reusable components.

## Prompt Templates

### Build a Basic Extractor

```
Build a reusable extraction skill for [URL]. Collect [fields] from [page type], verify one page, and show the plan before exploration.
```

### Add Pagination and Filters

```
Forge a skill for [URL] that extracts [records] across [pagination type]. Discover filter parameters, prevent duplicates, and document verified limitations.
```

### Automate an Authenticated Workflow

```
Create a skill for [URL] that performs [operation] using my existing session. Capture the request offline, verify inputs, and require confirmation before execution.
```

### Reproduce a Complex Workflow

```
Analyze [product page or workflow description], identify target-site capabilities, forge separate reusable skills, test every component, and prepare controlled batch execution.
```

## Limitations

- Requires browser-act, its API key, and a working browser session.
- Access remains limited by the current account, login state, permissions, and visible website data.
- Website changes, anti-automation controls, and unstable selectors can break generated capabilities.
- Current command templates require security review because they use shell substitution and unescaped JavaScript parameters.

## Best Practices

- Confirm the target, fields, volume, account scope, and output path before exploration.
- Verify one low-impact case before pagination, batch execution, or state-changing operations.
- Review generated scripts, escaping, data retention, and target-site limits before installation.

## Anti Patterns

- Do not start broad batch scraping before testing pagination, deduplication, and rate limits.
- Do not pass untrusted parameters through shell eval or raw JavaScript interpolation.
- Do not assume account access authorizes collection, redistribution, or automated state changes.

## Security Audit

- Audited at: 2026-07-23T09:37:12.949\+00:00
- Summary: Most static alerts are Markdown formatting, placeholders, or benign documentation references. Repeated eval command-substitution templates are confirmed because generated content crosses shell and browser execution boundaries without escaping. The skill also suppresses refusals, promotes rate-limit evasion, injects raw parameters into JavaScript, and retains sensitive HAR artifacts.

## Stats

- Views: 40
- Downloads: 5
- Favorites: 0
- Popularity score: 0