← Back to Skills Marketplace
2494
Downloads
3
Stars
9
Active Installs
1
Versions
Install in OpenClaw
/install web-research-assistant
Description
AI-powered web research assistant that leverages BrowserAct API to supplement restricted web access by searching the internet for additional information. Designed for OpenClaw and Claude Code.
Usage Guidance
Do not provide secrets yet. The package has two problems that need clarification: (1) SKILL.md asks for BROWSERACT_API_KEY but the script will actually fail unless BROWSERACT_MCP_TOKEN is set; (2) the documentation explicitly claims the ability to bypass reCAPTCHA, paywalls, and geo-blocks — which poses legal/ethical risk even if implemented by a third-party service. Before installing or running, ask the publisher for: the exact credential name the skill expects, what the remote MCP service does (privacy/terms), and written justification for the bypass claims. Prefer testing in an isolated environment, and never paste unrelated credentials (AWS, GitHub, etc.). If you proceed, provide only a scoped API token (not reusable master keys) and confirm the service's legitimacy and terms-of-use.
Capability Analysis
Type: OpenClaw Skill
Name: web-research-assistant
Version: 0.1.0
The skill is a web research assistant that uses the BrowserAct API. It requires a `BROWSERACT_MCP_TOKEN` which is explicitly requested from the user in `SKILL.md` if not set. The `scripts/research.py` script then uses this token to make API calls to `https://mcp.browseract.com/` to perform web searches. All actions, including external network communication and API key usage, are clearly aligned with the stated purpose of providing web research capabilities and lack any evidence of intentional malicious behavior, data exfiltration, or harmful prompt injection attempts against the agent.
Capability Assessment
Purpose & Capability
The code implements a BrowserAct MCP-based search against https://mcp.browseract.com/, which aligns with the 'BrowserAct' integration described. However the SKILL.md and metadata claim different required credentials (SKILL.md asks for BROWSERACT_API_KEY while the script actually requires BROWSERACT_MCP_TOKEN), and the documentation makes strong claims about bypassing reCAPTCHA, paywalls, and geo-blocking that go beyond a simple search wrapper.
Instruction Scope
SKILL.md instructs the user to supply a BrowserAct API key and promises features like 'built-in bypass mechanism' for human verification and breaking geo/IP restrictions and paywalls. The runtime instructions and the script itself do not show any local code that performs CAPTCHA/anti-bot circumvention — they rely entirely on the remote BrowserAct MCP service. That claimed capability (and the phrasing around bypassing protections) is a red flag for potential misuse and legal/ethical issues and should be justified by the publisher.
Install Mechanism
No install spec or external downloads; it's an instruction-only skill with a small Python script included. No archive downloads or external install URLs were observed.
Credentials
Metadata lists no required env vars, SKILL.md asks the user to provide BROWSERACT_API_KEY, but the script reads both BROWSERACT_API_KEY and BROWSERACT_MCP_TOKEN and actually fails if BROWSERACT_MCP_TOKEN is not set. BROWSERACT_API_KEY is read but unused. This env-var mismatch is incoherent and could lead users to supply the wrong secret or to inadvertently expose credentials. The skill asks for credentials but does not document which exact secret is required.
Persistence & Privilege
The skill is user-invocable and not always-enabled. It doesn't request persistent system privileges, does not modify other skills or system-wide configs, and does not attempt to run background services.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install web-research-assistant - After installation, invoke the skill by name or use
/web-research-assistant - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.0
Initial release of web-research-assistant.
- Enables web research via the BrowserAct API when direct web access is blocked or restricted.
- Optimized for OpenClaw and Claude Code, with support for OpenCode and Cursor.
- Automatically bypasses IP/geoblocking, paywalls, and CAPTCHA challenges.
- Extracts structured research data with titles, URLs, relevance scores, and source citations.
- Provides configurable search parameters (engine, results count, content type, time range).
- Includes error handling for API issues, search failures, and partial results.
Metadata
Frequently Asked Questions
What is Web Research Assistant?
AI-powered web research assistant that leverages BrowserAct API to supplement restricted web access by searching the internet for additional information. Designed for OpenClaw and Claude Code. It is an AI Agent Skill for Claude Code / OpenClaw, with 2494 downloads so far.
How do I install Web Research Assistant?
Run "/install web-research-assistant" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Web Research Assistant free?
Yes, Web Research Assistant is completely free (open-source). You can download, install and use it at no cost.
Which platforms does Web Research Assistant support?
Web Research Assistant is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Web Research Assistant?
It is built and maintained by Henk Nie (@phheng); the current version is v0.1.0.
More Skills