Claude Skill MartBrowse skillsWhat is a Skill?
 
← Back to list

browser-act Skill Forge (Reusable Scraper Builder)

Explores any website once to find its internal APIs (or DOM fallback), then auto-generates a verified, reusable Skill package — SKILL.md plus Python scripts — so you never have to re-explore.

AutomationAdvanced★ 5,316⑂ 253AI score 6/10Last updated: Aug 5, 2026

What it does

Tell it what data you need or what action you want automated on a given site, and it runs a full pipeline:

  • Requirements analysis: pins down the objective, target site, and whether you want immediate execution. If you give no URL, it actively searches and recommends 1–5 candidate sites with pros/cons.
  • Exploration (Phase 2): inspects network traffic to find the internal endpoints the site's own frontend calls; falls back to DOM extraction with strict selector-priority, Shadow DOM, and batch-validation rules when no API is viable.
  • Skill generation (Phase 3): wraps each verified call into a parameterized Python script (argparse) plus a SKILL.md, then runs end-to-end verification including error-path checks.
  • Automated testing, install, execute: a Sub-Agent runs the generated Skill against auto-designed test cases, fixes are applied until it passes, the Skill is installed, and your original bulk task is then executed.

The core promise is "explore once, reuse forever" — later runs skip discovery cost entirely, which is what makes large-scale extraction reliable.

Who it's for

  • Anyone repeating the same manual copy-paste routine on a website every day
  • Researchers and marketers needing hundreds to thousands of deduplicated records
  • Developers who want to understand how a site loads its data
  • Teams that want a durable, shareable tool instead of one-off browser automation

Prerequisites: the browser-act skill installed and a BrowserAct API key configured.

Example uses

  1. Build a news collector — "Make a skill that grabs the front-page posts from news.ycombinator.com" → discovers the list endpoint, emits scripts/list-posts.py + SKILL.md, tests it, installs it.
  2. Bulk listing pull — "Pull 2,000 listings from this real-estate site, no duplicates" → first forges a Skill with verified pagination/filter params, then writes a batch script that calls it at scale.
  3. Competitor price monitoring — Say only "track competitor price changes" with no URL; it researches candidate sources, lets you pick, and produces a reusable extraction Skill you can schedule daily.

· · · Install guide · · ·

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the browser-act-skill-forge folder from the GitHub repo browser-act/skills into my ~/.claude/skills/browser-act-skill-forge/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/browser-act/skills.git /tmp/browser-act-skills && cp -r /tmp/browser-act-skills/browser-act-skill-forge ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open your terminal.
  2. Clone the repository: git clone https://github.com/browser-act/skills.git /tmp/browser-act-skills
  3. Make sure the skills folder exists: mkdir -p ~/.claude/skills
  4. Copy this skill: cp -r /tmp/browser-act-skills/browser-act-skill-forge ~/.claude/skills/
  5. Also copy the base dependency: cp -r /tmp/browser-act-skills/browser-act ~/.claude/skills/ (if present in the repo).
  6. Sign up at https://www.browseract.com, get an API key, and configure it as instructed by the browser-act skill (environment variable or config file).
  7. Restart Claude Code and confirm browser-act-skill-forge appears in your skill list.
  8. Try a prompt like "Build a skill that extracts X from site Y". It will present a plan — approve it, and it runs exploration, generation, testing, and installation automatically.
  9. Generated packages land in output/{skill-name}/ under your current working directory by default.
View source on GitHub ↗License: MIT