Google vs Bing Search: Which to Use with Claude Code

AgentPatch provides both Google Search and Bing Search as MCP tools. For Claude Code, both are available once you connect — but they’re not identical. Here’s how to think about which one Claude Code will reach for, and when it matters.

Why This Matters

In practice, Claude Code decides which search tool to call based on the query. If both are available, it can use whichever it judges to be more appropriate. That’s usually fine, but it helps to understand what each engine is better at so you can guide Claude Code if needed.

Both tools cost 50 credits per call. The technical interface is the same. The difference is in what gets returned.

How They Compare

For coding and technical work specifically:

Google Search is the better default. Developer documentation, GitHub issues, Stack Overflow, technical blog posts, and package registries all tend to rank better on Google. If Claude Code is looking up a framework error, a library changelog, or a protocol spec, Google will return higher-quality results.

Bing Search is worth considering if:

  • You want a secondary search to cross-reference something
  • Google’s results for a specific query seem thin or low-quality
  • You’re dealing with non-English technical content where Bing’s index differs

For most Claude Code sessions, you won’t notice a difference because the queries are technical and Google dominates that space. But having both available means Claude Code can fall back to Bing when it’s useful.

Setup

Install the AgentPatch CLI (zero dependencies, Python 3.10+):

pip install agentpatch

Set your API key:

export AGENTPATCH_API_KEY=your_api_key

Then use it:

ap search "web search"
ap run agentpatch google-search --input '{"query": "test"}'

Get your API key from the AgentPatch dashboard.

Run this command to add AgentPatch as an MCP server:

claude mcp add -s user --transport http agentpatch https://agentpatch.ai/mcp \
  --header "Authorization: Bearer YOUR_API_KEY"

Replace YOUR_API_KEY with your actual key from the AgentPatch dashboard. Claude Code discovers all AgentPatch tools automatically.

Example

You ask Claude Code:

“I’m getting a ‘Cannot read properties of undefined’ error in my Vite build — only in production, not in dev. What’s usually causing this?”

Claude Code searches Google first — this is a developer-specific question where Google’s technical results will be more useful:

Searching Google for "Vite production build 'cannot read properties of undefined' SSR..."

This is a common pattern with Vite SSR builds. The most likely cause is accessing
browser-only globals (window, document, localStorage) during server-side rendering.
In dev mode, Vite polyfills some of these; in production it doesn't.

Check for any imports or code that references browser globals at the module level —
they need to be wrapped in `if (typeof window !== 'undefined')` or moved inside
component lifecycle hooks.

Want me to scan your codebase for potential violations?

Google’s developer-focused results led to a clear diagnosis. Bing would likely return similar results here, but Google’s index of Stack Overflow and framework docs tends to be more current.

Wrapping Up

For Claude Code, Google Search is the right default for technical queries. Bing Search is a useful backup, and having both available costs nothing extra once you’ve connected AgentPatch. The same connection also gives access to image search, maps, and the full marketplace. More at agentpatch.ai.

← Back to blog