Using Web Search

Using Web Search & Docs in Cascade

Cascade can access information beyond your codebase and its training data by utilizing web search and curated documentation. This provides real-time context relevant to your prompts.

• • Admin Control: For enterprise security, administrators can toggle web search and docs access on or off.

@docs: Accessing Curated Documentation

• • Purpose: Quickly access context from pre-selected, high-confidence documents for various software applications and libraries.
• • How to Use:
  1. 1. Type @docs in Cascade.
  2. 2. Search for and select the relevant document (e.g., "Vercel docs").
  3. 3. Prompt Cascade with your task (e.g., "build a plan for implementing Vercel auth").
• • Process: Cascade will search through the selected documentation to understand and plan the requested changes.

@web: Explicit Web Search

• • Purpose: Explicitly instruct Cascade to use public web search to answer questions or gather information.
• • How to Use:
  1. 1. Type @web followed by your query.
  • • Example: "@web What is the most recent version of JavaScript?"
• • Process: Cascade performs a search query and scrapes relevant web pages (e.g., Wikipedia, official documentation) to provide up-to-date answers.

Providing URLs Directly

• • Purpose: Have Cascade analyze and understand the content of a specific public web page.
• • How to Use:
  1. 1. Paste a public URL directly into Cascade.
  2. 2. Prompt Cascade with your task related to the URL's content.
  • • Example: "Analyze this page [link to Pandas pivot_table docs] and explain how to use pivot_table."
• • Process: Cascade navigates the provided link, even drilling down into specific sections if necessary, to gather information and respond to your prompt.

These features extend Cascade's capabilities, allowing it to leverage real-time, external information to assist you. For more on how Cascade connects to external tools, see our content on Model Context Protocol (MCP).

• • "Find the official documentation for this API"

Examples of Web Search Queries

Documentation Lookup:

• • "Find the latest React 18 hooks documentation"
• • "Show me AWS Lambda Python runtime configuration options"

Troubleshooting:

• • "Search for solutions to this specific error message"
• • "Find recent discussions about this package compatibility issue"

Learning:

• • "Find tutorials on implementing GraphQL with TypeScript"
• • "Search for current best practices in microservices architecture"

Web Search Features

Quick Start: Simply ask questions and Cascade will search when needed.

Source Citations: Cascade shows you where information came from.

Multiple Sources: Combines information from various authoritative sources.

Fresh Information: Gets the most current available information.

Best Practices

• • Ask specific, focused questions rather than broad topics
• • Include your technology stack and version numbers
• • Request official documentation when available
• • Verify critical information from multiple sources
• • Use web search for time-sensitive or rapidly changing topics