browser-act-skill-forge

name

browser-act-skill-forge

description

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', 'forging a skill'. Also applies to: need to regularly extract data from a site, build a custom scraper, create a reusable web data tool, generate a social media monitoring skill, turn manual browser tasks into automated skills.

BrowserAct Skill Forge

Turns any website's data extraction or operation needs into Agent-callable capabilities. User describes what data to get or what action to perform on which site; this Skill automatically explores implementation paths (API endpoints first, DOM fallback when no API), then generates deploy-ready Skill packages (SKILL.md + Python scripts) after verification. Explore once, reuse forever.

Install Prerequisites

Make sure browser-act is installed and working first:

uv tool install browser-act-cli --python 3.12
browser-act --version

The Skill Forge Workflow

Phase 1 (Requirements) → Phase 2 (Explore Site/API) → Phase 3 (Generate Skill) → Delivery (Test + Install + Execute)

Phase 1 — Requirements Analysis

1. Identify: core objective, target site, execution intent, output directory
2. For vague requests → research candidate sites via web search
3. Decompose into independent capabilities
4. Confirm execution plan with user

Phase 2 — Capability Exploration

Read the exploration reference based on type:

• Extraction → API endpoints first (fetch reproduction), then UI trigger + Network capture, then DOM extraction
• Operation → UI interaction flow mapping

Priority: API endpoints > Network capture > DOM extraction > AI workflow

Phase 3 — Skill Generation

For each verified capability:

1. Encapsulate JS into scripts/{feature-name}.py (argparse + f-string JS)
2. Verify end-to-end: python script.py {params} → eval "$(python script.py {params})"
3. Generate SKILL.md per template with capability components, enum parameters, pagination

Delivery Flow

1. Automated testing of all components
2. Install the Skill
3. Report results to user
4. Execute the user's original task (if execution intent was identified)

Key Principles

• Operational boundary = what user can manually do in browser (copy-paste equivalence)
• All data stays local — traffic inspection, HAR, extraction on user's machine
• Data access through browser only — never through third-party APIs or scraping platforms
• Partial success counts as success — inform user about uncovered parts
• Exploration cap: 100 tool call steps maximum
• Error handling: deterministic failure = one attempt; transient failure = one retry

Output Structure

output/{skill-name}/{site-slug}-{capability-slug}/
├── SKILL.md
└── scripts/
    └── {feature-name}.py

Experience Notes

Path: {working-directory}/browser-act-skill-forge-memories/forge.memory.md

Before execution: If the file exists, read it first.

After execution: Append unexpected situations and conclusions.