Skip to content

Web Tools

7 tools for web search and URL fetching.

Search the web using Tavily API.

tavily_search(query: str, max_results: int = 5) -> str

Examples:

> search for "Python async best practices"
> find tutorials on FastAPI

API Key Required

Set TAVILY_API_KEY in your .env file. Get a free key at tavily.com

Search using DuckDuckGo (no API key needed).

duckduckgo_search(query: str, max_results: int = 5) -> str

Example:

> search DuckDuckGo for "Python logging"

jina_reader

Extract clean content from a webpage.

jina_reader(url: str) -> str

Example:

> read the content of https://example.com/article

Returns: Clean, formatted text content.

fetch_url_content

Fetch raw content from any URL.

fetch_url_content(url: str) -> str

Example:

> fetch https://api.example.com/data.json

Search Wikipedia articles.

wikipedia_search(query: str) -> str

Example:

> search Wikipedia for "Machine Learning"

Search arXiv for academic papers.

arxiv_search(query: str, max_results: int = 5) -> str

Example:

> find papers on "transformer architecture"

Search GitHub repositories and code.

github_search(query: str, search_type: str = "repositories") -> str

Search Types:

  • repositories - Find repos
  • code - Find code snippets

Example:

> search GitHub for Python CLI frameworks

Fallback Behavior

QUASAR tries multiple web tools in sequence:

1. Tavily (if API key)
2. DuckDuckGo
3. Jina Reader

If one fails, it automatically tries the next.