browser-commerce-base

Browser-first workflow for e-commerce skills on OpenClaw. Use when shopping, comparing products, extracting prices, checking seller/store details, reading logged-in carts/orders, validating coupons, checking delivery availability, or collecting screenshots from Taobao, JD, Pinduoduo, Meituan, Eleme, VIP, SHEIN, Dianping, and similar commerce platforms.

Safety Notice

This listing is from the official public ClawHub registry. Review SKILL.md and referenced scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "browser-commerce-base" with this command: npx skills add harrylabsj/browser-e-commerce

Browser e-Commerce

Use this skill as the shared browser workflow for e-commerce tasks.

Profile selection

Choose browser profile by risk and need:

  • openclaw: default for public browsing, search results, product pages, screenshots, and cross-platform comparison.
  • user: use only when existing login state matters, such as carts, coupons, orders, addresses, or member prices.
  • chrome-relay: use only when the user explicitly wants the extension attach-tab flow.

Default rule:

  • public commerce pages → openclaw
  • logged-in private pages → user

Base workflow

  1. Start with browser status and browser start if needed.
  2. Open/focus the target site in a stable tab.
  3. Run snapshot before any action.
  4. Prefer snapshot --interactive / actionable refs over brittle CSS assumptions.
  5. After navigation, filter changes, popups, or SKU changes, re-run snapshot.
  6. Use highlight when a button/card/price block is ambiguous.
  7. Capture evidence with screenshots when the result matters.
  8. When the page is highly dynamic, check requests, errors, and response body.

Commerce extraction order

For most shopping tasks, extract in this order:

  1. platform
  2. page type: search result / product detail / cart / order / coupon / store
  3. title
  4. current price
  5. final price / coupon price / member price
  6. store or seller
  7. sales / rating / review count
  8. shipping or delivery promise
  9. SKU / spec options
  10. risks / caveats
  11. link + screenshot evidence

Standard page workflows

Search results page

Goal: compare multiple candidates quickly.

Do:

  • open search page
  • snapshot interactive elements
  • identify product cards, price text, store names, tags, sales, coupons
  • collect top candidates into structured notes
  • screenshot result page if ranking matters

Product detail page

Goal: make one item decision-ready.

Do:

  • extract title, price, final price, store, shipping, service badges
  • inspect SKU/spec selectors
  • note coupons, subsidies, timed promotions, member-only prices
  • if specs are changed, re-snapshot before re-reading price
  • screenshot detail evidence when recommending a purchase

Cart / order / coupon page

Goal: use private login state safely.

Do:

  • switch to profile="user" only if needed
  • prefer read/check actions over side effects
  • stop before irreversible actions such as payment or final order submission
  • summarize discounts, totals, coupon applicability, and delivery info

Location-sensitive commerce page

Goal: judge delivery or stock by location.

Do:

  • use geolocation or user-browser context when available
  • check whether delivery area, ETA, store availability, or fee changes by location
  • state clearly when output depends on location assumptions

Safety boundaries

  • Do not complete payment.
  • Do not place final orders without explicit confirmation.
  • Do not change account settings, addresses, or payment methods unless explicitly asked.
  • For logged-in pages, prefer observation and recommendation over action.

Debug workflow

When automation becomes unstable:

  1. re-run snapshot
  2. highlight the target ref
  3. inspect errors
  4. inspect requests
  5. use response body for commerce APIs when visible UI is incomplete
  6. use trace only when needed

Read references as needed

  • Read references/browser-commerce-schema.md for the reusable output schema.
  • Read references/platform-adaptation.md for platform-specific browser strategy.

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

Automation

PolyFly Predictions

Prediction markets for AI agents on Hedera. Create markets, place bets, earn HBAR. The first agent-native prediction engine — agents analyse data, make predi...

Registry SourceRecently Updated
Automation

Wip Universal Installer

Reference installer for agent-native software. Scans a repo, detects which interfaces it exposes, and installs them all.

Registry SourceRecently Updated
Automation

Moltline

Public topics and posts plus private XMTP messaging for agents

Registry SourceRecently Updated
Automation

dchat

Decentralized P2P bot-to-bot messaging over NKN. Send and receive text, images, audio, and files without any centralized server. Private, encrypted, serverless.

Registry SourceRecently Updated