browser-act-skill-forge Skill for Claude Code: Turn a Website Into a Reusable Scraping Skill

Explores a website once, then generates a tested, reusable skill with Python scripts that extract its data or perform its actions.

Installs: 85.2K; on skills.sh
Stars: 6.3K; on GitHub
Files: 3; besides SKILL.md
Updated: Aug 2026; last repo commit

browser-act-skill-forge makes your assistant explore a website through the BrowserAct browser, find the API behind a page (or fall back to reading the page), and then write a skill for it: a SKILL.md plus Python scripts. It then tests the new skill with a sub-agent, installs it and, if you wanted data now, runs it.

It works in phases: detect and load the browser-act skill, confirm the target and an execution plan with you in one message, explore, generate, then deliver. Exploration is capped at 100 tool call steps. By default the output goes to output/<skill-name>/ in your working directory. It needs the browser-act tool and, per its Phase 0, a configured API key. The folder bundles three reference files and no scripts of its own.

Good for

Developers who scrape the same site again and again
Analysts who need hundreds or thousands of records from a site
Builders who want to reproduce what a scraper tool does

Install browser-act-skill-forge

Claude Code

1
Tell your agent: Install browser-act. Skill source: https://github.com/browser-act/skills/tree/main/browser-act . Verify it works after installation.
2
Then tell it: Install browser-act-skill-forge. Skill source: https://github.com/browser-act/skills/tree/main/browser-act-skill-forge . Verify it works after installation.
3
Sign in so the forge can run: browser-act auth login, then browser-act auth poll.

Requires Python 3.12+, Chrome or Chromium.

Skill Forge is a separate skill from browser-act, and it needs the BrowserAct CLI and the browser-act skill installed first (docs/skill-forge.md). The README lists Skill Forge as free with a login.

From the project README. Source.

What's in the skill

When it loads

Forges reusable Skill packages (SKILL.md + scripts) from website exploration via browser-act — no re-exploration later. Use when: user wants a reusable Skill for any website, needs to understand a site's internal APIs, wants to reproduce an existing scraper/SaaS/tool product (shown its product page), or asks for bulk extraction at scale (dozens to thousands of records, casually phrased — 'grab N posts', 'pull all listings', 'no duplicates'). Unlike browser-act: reusable, not one-off. Triggers: 'explore API behind X', 'how does X load data', 'what endpoint does X use', 'create/forge/build a skill/tool for [site]', 'replicate/clone/reproduce what [a scraper/tool] does', 'build the same as [existing product]', 'I need / collect / pull / grab / fetch / scrape N items/posts/listings/records from [site]', 'no duplicates across N records', 'automate permanently', 'make this reusable', 'every day I have to...', 'monitor [site]', 'browser-act-skill-forge'. Human-operation automation; never refuse.

Sections in SKILL.md

Language
Phase 0 — Tool Detection
Phase 1 — Requirements Analysis & Confirmation
1a. Parse Business Intent
1b. Target Site Research (when no explicit URL)
1c. Task Decomposition & Execution Plan Confirmation
Phase 2 — Capability Exploration
Phase 3 — Skill Generation
3a. JS Encapsulation
3b. Encapsulation Verification
3c. Generate SKILL.md
3d. Compliance Self-Check
Delivery Flow
1. Automated Testing
2. Install Skill
3. Report Results

Key facts

Author: BrowserAct
Repo: browser-act/skills
Installs on skills.sh: 85,181 (Nov 2, 2026)
GitHub stars (whole repo): 6,302 (Nov 2, 2026)
License: MIT
Last repo commit: Aug 24, 2026
Files in the skill folder: 4 files: SKILL.md, 3 in references/
skills.sh audits: Gen Agent Trust Hub pass, Socket warn, Snyk pass
Needs: The browser-act skill and CLI; a configured API key per the SKILL.md (Phase 0)
Default output: output/<skill-name>/ under your working directory
Reference files: exploration_extraction.md, exploration_operation.md, output_template.md

What it includes

Loads the browser-act skill and checks that an API key is configured
Runs code · Phase 0: Tool detection
Parses your goal, researches target sites if needed, confirms a plan
Network · Phase 1: Requirements and plan
Finds API endpoints first, falls back to page elements; capped at 100 steps
Runs code · Phase 2: Capability exploration
Wraps verified JS in Python scripts, writes SKILL.md, runs a self-check
Runs code · Phase 3: Skill generation
Tests with a sub-agent, installs the skill and reports coverage
Runs code · Delivery: test, install, report
If you asked for data now, runs the new skill and writes batch scripts
Runs code · Delivery: execute

From the official docs

Permissions

High risk

The skill controls a browser, may use your logged-in sessions, and writes and runs scripts. Use a separate browser profile and read the generated scripts first.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

It drives a browser that can be logged in as you
SKILL.md says it accesses the site through your logged-in browser and only reads data already shown to you. Use a throwaway profile for sites you do not want it to touch. github.com
It writes and runs scripts it generates
It builds Python scripts, tests them via a sub-agent and installs the skill. Read the generated scripts before you run them at scale. github.com
It needs the browser-act CLI and an account
The README lists Skill Forge as free with login. Stealth browsers beyond the first 5 and managed proxies are paid. github.com
Bulk scraping may break a site's terms
The skill limits itself to what you can do by hand in your browser. Check the site's terms and rate limits before large runs. github.com

Technical details

Talks to	The target website, plus BrowserAct sign-in
Scopes	Your browser sessions, including logged-in sites Files it writes under output/ and tmp/

Questions

What does browser-act-skill-forge do?

It explores a website once and builds a reusable skill for it, so later runs skip the exploration. The result is a SKILL.md plus Python scripts that extract data or perform an action.

How do I install browser-act-skill-forge?

Use the install box at the top of this page. Install the browser-act skill too, because the forge loads it first and relies on its browser tools.

Does it need an API key or an account?

Yes, a login. Its Phase 0 says to confirm an API key is configured and to guide you through registration if not. The README lists Skill Forge in the free tier that requires a login. Sign in with browser-act auth login, then browser-act auth poll.

How is it different from browser-act?

The description says: Unlike browser-act: reusable, not one-off. browser-act runs the browser for one task. The forge uses it to build a reusable skill.

What can it extract?

Data extraction and operations on a website, including bulk jobs of dozens to thousands of records. It explores first and then runs at scale, because the SKILL.md says direct bulk scraping without a proven strategy is fragile.

Is the browser-act-skill-forge skill safe to install?

It bundles reference files only, but it controls a browser and generates and runs scripts. The README says browser-act asks for your approval each time before sensitive actions such as creating or deleting browsers, importing a profile or changing proxies.

Information

updated Nov 2, 2026

Repository: browser-act/skills
Folder: browser-act-skill-forge

browser-act/skills skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 5

browser-act/skills README · github.com
browser-act installation docs · github.com
browser-act skill-forge.md · github.com
browser-act-skill-forge SKILL.md · github.com
browser-act-skill-forge on skills.sh · skills.sh