← Back to Skills Marketplace
whjstc

Browsecraft

by RoyWu · GitHub ↗ · v0.1.3 · MIT-0
cross-platform ✓ Security Clean
262
Downloads
0
Stars
0
Active Installs
4
Versions
Install in OpenClaw
/install browsecraft
Description
Memory-oriented browser automation skill for repeatable web workflows (login, extraction, bulk actions, form filling, screenshots, checks) across RoxyBrowser...
README (SKILL.md)

BrowseCraft Skill

Installation Prerequisites

  1. Install CLI: npm install -g browsecraft-cli
  2. Verify install: browsecraft --help
  3. Optional (RoxyBrowser only): configure your Roxy API endpoint and token in local env/config.

If the CLI is missing, stop and ask the user to install it first. Do not assume local package scripts are available.

Recommended Flow

  1. Check status: browsecraft status
  2. Start browser if needed: browsecraft start
  3. Open target page: browsecraft open \x3Curl>
  4. Capture snapshot: browsecraft snapshot
  5. Prefer click-ref / fill-ref for stable interactions
  6. Re-snapshot after page transitions
  7. Capture result evidence: browsecraft screenshot

Backend Strategy

  • RoxyBrowser: browsecraft start --type roxy --roxy-window-id ...
  • Camoufox: browsecraft start --type camoufox (optional --camoufox-path)
  • Existing endpoint: browsecraft connect \x3Cendpoint> --type \x3Cchrome|camoufox|roxy>

Credential & Scope Policy

  • Roxy credentials are optional and only needed when the user explicitly chooses RoxyBrowser.
  • Prefer environment variables over CLI token flags. Avoid printing or echoing secrets.
  • Never send credentials or page data to third-party endpoints unless the user explicitly requests it.
  • If a step needs network access outside the target website, ask for confirmation first.

Stability Rules

  • If element lookup fails: refresh with snapshot, then retry.
  • If page is unstable: use wait-for before interaction.
  • Always return structured output: objective / steps / result / failure reason / next action.
Usage Guidance
This skill appears coherent for a CLI-based browser automation tool, but before installing: 1) verify the npm package (browsecraft-cli) on the npm registry and inspect its homepage/repository and maintainer — the registry metadata provided here lacks a source URL; 2) avoid global installs until you confirm the package owner and version; 3) prefer entering sensitive tokens via environment variables (not CLI flags) and only provide Roxy tokens when needed; 4) when the agent asks to access endpoints outside the target site, confirm interactively; 5) consider testing the CLI in an isolated environment (sandbox/container) or reviewing the package code before trusting it in production. If you want a higher-confidence recommendation, provide the npm package link or repository so its code and maintainers can be inspected.
Capability Analysis
Type: OpenClaw Skill Name: browsecraft Version: 0.1.3 The BrowseCraft skill is a legitimate browser automation tool designed for workflows like form filling and data extraction using the 'browsecraft-cli' NPM package. The SKILL.md file includes explicit safety policies regarding credential handling and network access, and the requested Bash permissions are restricted to the 'browsecraft' command namespace, showing no signs of malicious intent or data exfiltration.
Capability Assessment
Purpose & Capability
Name and description describe browser automation and the skill only requires a 'browsecraft' CLI binary and documents commands that match that purpose. The declared npm package (browsecraft-cli) corresponds to the required binary, so required pieces are proportional to the stated capability.
Instruction Scope
SKILL.md only instructs using the browsecraft CLI (start/open/snapshot/screenshot/etc.), optional configuration for RoxyBrowser, and good-secret-handling guidance. It does not instruct reading unrelated system files or sending data to external endpoints without explicit user consent.
Install Mechanism
Install is via npm (browsecraft-cli) which is a reasonable packaging method for a CLI. However, the registry metadata lists no source/homepage and no code files are included for review — installing an npm package runs third-party code and has moderate risk if the package or owner are untrusted.
Credentials
No required environment variables are declared. Roxy credentials are mentioned as optional and only for Roxy usage; the skill explicitly recommends using environment variables and avoiding printing secrets, which is proportionate to its functionality.
Persistence & Privilege
Skill is not forced-always, does not request system-wide config or access to other skills, and follows the normal model-invocation defaults. No elevated persistence or privileges are requested.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install browsecraft
  3. After installation, invoke the skill by name or use /browsecraft
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.3
- Added Openclaw metadata to SKILL.md for improved package management and installation instructions. - Clarified RoxyBrowser installation by changing environment variable setup to general configuration in local environment or config file. - Revised description of credential policy to refer to "Roxy credentials" instead of specific environment variables. - General cleanup and improvements to installation and backend strategy instructions.
v0.1.2
- Added installation prerequisites and instructions for the browsecraft CLI. - Updated backend credential handling to prefer environment variables and clarified RoxyBrowser usage. - Introduced a credential & scope policy section with security recommendations. - Removed explicit Bash tool path restrictions, now allowing Bash(browsecraft:*). - Clarified to confirm with the user before any network access outside the target website.
v0.1.1
- Removed the reference and link to the Chinese documentation in SKILL.md. - No changes to core functionality or usage instructions.
v0.1.0
Initial release of BrowseCraft: A memory-oriented browser automation tool. - Supports repeatable web workflows: login, extraction, bulk actions, form filling, screenshots, and checks. - Works with RoxyBrowser, Camoufox, and Chrome via flexible backend selection. - Includes commands for browser status, startup, page opening, snapshots, and screenshots. - Designed for workflow stability: element re-checks, waiting for elements, structured output, and recovery from failures.
Metadata
Slug browsecraft
Version 0.1.3
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 4
Frequently Asked Questions

What is Browsecraft?

Memory-oriented browser automation skill for repeatable web workflows (login, extraction, bulk actions, form filling, screenshots, checks) across RoxyBrowser... It is an AI Agent Skill for Claude Code / OpenClaw, with 262 downloads so far.

How do I install Browsecraft?

Run "/install browsecraft" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Browsecraft free?

Yes, Browsecraft is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Browsecraft support?

Browsecraft is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Browsecraft?

It is built and maintained by RoyWu (@whjstc); the current version is v0.1.3.

💬 Comments