Web Research Assistant
AdvisoryAudited by Static analysis on Apr 30, 2026.
Overview
No suspicious patterns detected.
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The agent could be steered toward accessing or extracting content in ways that violate website controls, terms, or legal expectations.
The skill explicitly tells users to use it in paywall and human-verification scenarios, which can encourage bypassing website access controls rather than only researching public, authorized sources.
“Paywalls prevent data extraction” and “内置绕过机制，无需处理 reCAPTCHA 或其他验证挑战。”
Use only for public or otherwise authorized research. The skill should remove paywall/CAPTCHA-bypass positioning and require explicit user confirmation before attempting restricted-source research.
Users may over-trust generated reports or assume extracted facts are verified when the code only performs lightweight snippet extraction and recommends later verification.
The documentation makes an absolute no-hallucination and precision guarantee for web extraction, which overstates reliability for search snippets and third-party web data.
“没有幻觉，确保稳定精准提取数据”
Treat results as unverified research leads. The skill should replace absolute accuracy claims with clear limitations and validation requirements.
Users may be surprised by the required credential or may paste a secret into chat because the credential contract is inconsistent.
The executable requires a BrowserAct token, but registry metadata declares no required environment variables and SKILL.md asks for BROWSERACT_API_KEY instead.
BROWSERACT_MCP_TOKEN = os.environ.get("BROWSERACT_MCP_TOKEN"); ... if not BROWSERACT_MCP_TOKEN: print("Error: BROWSERACT_MCP_TOKEN environment variable not set")Declare the required credential in metadata, use one consistent environment variable name, and prefer setting it through a shell environment or secret manager rather than chat.
Research topics, and potentially sensitive query text, will be shared with BrowserAct.
The user’s query and search parameters are sent to the BrowserAct provider with a bearer token. This is expected for the integration, but it is an external data flow.
MCP_SERVER_URL = "https://mcp.browseract.com/"; ... requests.post(MCP_SERVER_URL, headers={"Authorization": f"Bearer {BROWSERACT_MCP_TOKEN}"}, json=payload, timeout=60000)Avoid sending confidential queries unless BrowserAct is approved for that data, and disclose this provider data flow clearly to users.