Web Browsing.Bak

Browse and summarize websites, extract content from URLs, search the web for information. Use when user asks to visit a website, get webpage content, search...

Security Scan

The skill's code and instructions match its stated purpose (web search and fetching pages), but there are small inconsistencies and missing declarations (undeclared Python dependencies and metadata mismatch) that warrant caution before installing.

Purpose & Capability

The name/description (browse, summarize, search) align with the included code and SKILL.md. The provided script implements search_web and fetch_url which perform DuckDuckGo searches and fetch arbitrary webpages — these are coherent with the skill's purpose. The skill does not request unrelated credentials or access.

Instruction Scope

Runtime instructions are specific: use web_search/fetch_url/extract_content for searches and fetching URLs. The SKILL.md does not instruct the agent to read unrelated files, harvest credentials, or transmit data to unexpected endpoints. It does direct the agent to fetch arbitrary user-supplied URLs and results from DuckDuckGo (expected for this purpose).

Install Mechanism

The skill is 'instruction-only' (no install spec) but includes a Python script that depends on third-party packages (requests, bs4). Those dependencies are not declared in the metadata or SKILL.md. This mismatch can cause runtime failures or surprise if the execution environment lacks these packages; it's not evidence of maliciousness but is an operational/integrity concern.

Credentials

No environment variables, credentials, or config paths are requested. The network access performed (HTTP GETs to DuckDuckGo and to user-provided URLs) is proportional to a web-browsing skill. There are no undeclared attempts to read local files or tokens.

Persistence & Privilege

The skill does not request persistent/always-on presence (always: false) and does not claim to modify other skills or system-wide settings. Normal autonomous invocation is allowed (disable-model-invocation: false), which is platform default and not, by itself, suspicious here.

What to consider before installing

What to consider before installing: - The skill will make HTTP requests to DuckDuckGo and any URLs you provide. Do not pass URLs that contain sensitive tokens or that require authentication unless you understand the risks. - The included Python script uses third-party libraries (requests, bs4) but the skill does not declare dependencies or provide an install step. Verify your environment has these packages or install them in a sandbox before running. - Metadata mismatch: _meta.json ownerId differs from the registry owner id and the package is listed as source: unknown with no homepage. This doesn't prove malice but reduces provenance and should make you more cautious. - Review the script before use (it is short and readable). Running it in an isolated environment (container or restricted VM) is recommended if you will fetch untrusted pages. - If you plan to have the agent fetch pages that require credentials or that are internal/private, do not provide secrets to this skill; it has no mechanism for safe credential handling. If you want to proceed, verify dependencies (requests, beautifulsoup4), and consider running the code in a sandboxed environment. If provenance is important, ask the publisher for source and homepage information or prefer a browsing tool from a known source.

Web Browsing Skill

This skill enables browsing websites, extracting content from URLs, and searching the web for information.

When to Use

Use this skill when:

• User asks you to visit a website (e.g., "Check out example.com")
• User wants webpage content summarized
• User needs current information that requires internet search
• User provides a URL and wants it analyzed
• User asks to search for something online

How It Works

1. Direct URL Access

When given a specific URL:

User: "What's on https://example.com?"
→ Visit the page, extract main content, summarize key points

2. Web Search

When user asks to search:

User: "Find information about climate change"
→ Perform web search, present top results with summaries

3. Content Extraction

For specific data extraction:

User: "Get the latest news from techcrunch.com"
→ Navigate to site, extract relevant articles/headlines

Tools Available

• web_search: Search the web for information
• fetch_url: Visit and retrieve webpage content
• extract_content: Parse HTML and extract structured data

Best Practices

1. Be specific - Tell me what you want from the page (summary, specific data, latest news)
2. Provide URLs - If you have a specific page, share the URL directly
3. Clarify intent - Let me know if you need:
   • Quick summary
   • Detailed analysis
   • Specific data points
   • Latest updates

Examples

✅ "Visit https://news.ycombinator.com and summarize today's top stories"
✅ "Search for the latest React.js tutorial"
✅ "Check what's on Wikipedia's page about quantum computing"
✅ "Find pricing information from apple.com/iphone"
❌ Just say "browse the web" - be more specific!

Limitations

• Cannot interact with JavaScript-heavy sites (may miss dynamic content)
• Some sites block automated access
• Video/audio content cannot be played, only described if available
• Login-required pages won't work without credentials

---

Ready to browse! Just give me a URL or tell me what to search for. 🌐