How long does this set up the scavio mcp server in claude code (5-minute guide) tutorial take?

Most developers complete this tutorial in 15 to 30 minutes. You will need a Scavio API key (free tier works) and a working Python or JavaScript environment.

What do I need before starting?

Claude Code installed and signed in (the same flow works in Cursor, VS Code, Windsurf, and ChatGPT).. A Scavio API key from the dashboard at scavio.dev — signup gives you 50 free credits, no card.. Network egress from the machine running Claude Code (the remote MCP lives at mcp.scavio.dev, so a firewall that blocks outbound HTTPS will break it).. A Scavio API key gives you 50 free credits on signup.

Can I run this tutorial with the free tier?

Yes. The free tier includes 50 credits on signup, which is more than enough to complete this tutorial and prototype a working solution.

What frameworks does this work with?

Scavio has a native LangChain package (langchain-scavio), an MCP server, and a plain REST API that works with any HTTP client. This tutorial uses the raw REST API, but you can adapt to your framework of choice.

Add Scavio MCP to Claude Code in 5 Minutes

Run `claude mcp add --transport http scavio https://mcp.scavio.dev/mcp --header "x-api-key: YOUR_API_KEY"`, then type `/mcp` inside Claude Code to confirm the tools loaded. That's the whole setup, and it takes under five minutes. After it connects, Claude can search Google, Reddit, YouTube, Amazon, and TikTok with live data instead of guessing from its training cutoff. The two things people trip on: the MCP server authenticates with an `x-api-key` header (NOT `Authorization: Bearer` — Bearer is only for the REST API), and you have to actually run `/mcp` to verify, because a silent failure looks identical to a working setup until Claude tries to call a tool.

Prerequisites

Claude Code installed and signed in (the same flow works in Cursor, VS Code, Windsurf, and ChatGPT).
A Scavio API key from the dashboard at scavio.dev — signup gives you 50 free credits, no card.
Network egress from the machine running Claude Code (the remote MCP lives at mcp.scavio.dev, so a firewall that blocks outbound HTTPS will break it).

Walkthrough

Step 1: Get your Scavio API key

Sign in at scavio.dev, open the dashboard, and copy your API key. A new account ships with 50 free credits, which is enough to test every tool a few times. Most calls cost 1 credit; Reddit post (full comment tree) and Google with light_request:false cost 2.

Bash

# Your key looks like this (copy it from the dashboard):
# sk_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxx

Step 2: Add the server with one command

The fastest path is the CLI. This registers the remote Scavio MCP over HTTP and attaches the auth header in one shot. Note the header is x-api-key, not Bearer.

Bash

claude mcp add --transport http scavio https://mcp.scavio.dev/mcp \
  --header "x-api-key: YOUR_API_KEY"

Step 3: Or edit the JSON config directly

If you'd rather configure by hand (or you're in Cursor/Windsurf/VS Code), add this block to your MCP config file. The mcpServers entry points at the remote URL and sets the same x-api-key header. This JSON is identical across every client.

JSON

{
  "mcpServers": {
    "scavio": {
      "url": "https://mcp.scavio.dev/mcp",
      "headers": {
        "x-api-key": "YOUR_API_KEY"
      }
    }
  }
}

Step 4: Verify with /mcp before you trust it

Open Claude Code and type /mcp. You should see scavio listed as connected, with its tools (Google, Reddit, YouTube, Amazon, TikTok search, and more). If it shows failed or connecting forever, it is almost always one of two things: a bad/expired key, or you pasted Bearer instead of x-api-key. Fix the header and re-add.

Bash

/mcp
# Expected: scavio  connected  (tools: search_google, reddit_search, youtube_search, ...)

Step 5: Ask Claude to search

Once /mcp shows connected, just ask in plain language. Claude picks the right Scavio tool and returns live results. No extra wiring.

Bash

# In Claude Code, type:
Search Google for "best web data API for AI agents 2026" and summarize the top 5 results.

Python Example

Python

# The MCP server is driven through Claude, not Python. But here's the equivalent
# CLI flow you'd run to add and inspect the server from a terminal:

# 1. Add the remote Scavio MCP (HTTP transport, x-api-key auth)
claude mcp add --transport http scavio https://mcp.scavio.dev/mcp \
  --header "x-api-key: YOUR_API_KEY"

# 2. List configured servers and confirm scavio is there
claude mcp list

# 3. Inside an interactive Claude Code session, verify the tools loaded
#    /mcp  ->  scavio  connected

JavaScript Example

JavaScript

// Same config, JSON form. Drop this into the mcpServers block of your
// Claude Code / Cursor / Windsurf / VS Code MCP settings file.
{
  "mcpServers": {
    "scavio": {
      "url": "https://mcp.scavio.dev/mcp",
      "headers": {
        "x-api-key": "YOUR_API_KEY"
      }
    }
  }
}

Expected Output

JSON

After `/mcp`, Claude Code shows:

  scavio  connected
  tools: search_google, reddit_search, reddit_post, youtube_search, youtube_metadata, amazon_product, walmart_product, tiktok_profile, tiktok_search_videos, tiktok_hashtag, ... (33 tools total)

Then, asking Claude to "search Google for X" returns live organic results, related searches, and (with the deeper mode) people-also-ask and knowledge-graph data — pulled fresh, not from the model's training cutoff.

Prerequisites

Claude Code installed and signed in (the same flow works in Cursor, VS Code, Windsurf, and ChatGPT).
A Scavio API key from the dashboard at scavio.dev — signup gives you 50 free credits, no card.
Network egress from the machine running Claude Code (the remote MCP lives at mcp.scavio.dev, so a firewall that blocks outbound HTTPS will break it).

Walkthrough

Step 1: Get your Scavio API key

Bash

# Your key looks like this (copy it from the dashboard):
# sk_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxx

Step 2: Add the server with one command

The fastest path is the CLI. This registers the remote Scavio MCP over HTTP and attaches the auth header in one shot. Note the header is x-api-key, not Bearer.

Bash

claude mcp add --transport http scavio https://mcp.scavio.dev/mcp \
  --header "x-api-key: YOUR_API_KEY"

Step 3: Or edit the JSON config directly

JSON

{
  "mcpServers": {
    "scavio": {
      "url": "https://mcp.scavio.dev/mcp",
      "headers": {
        "x-api-key": "YOUR_API_KEY"
      }
    }
  }
}

Step 4: Verify with /mcp before you trust it

Bash

/mcp
# Expected: scavio  connected  (tools: search_google, reddit_search, youtube_search, ...)

Step 5: Ask Claude to search

Once /mcp shows connected, just ask in plain language. Claude picks the right Scavio tool and returns live results. No extra wiring.

Bash

# In Claude Code, type:
Search Google for "best web data API for AI agents 2026" and summarize the top 5 results.

Python Example

Python

# The MCP server is driven through Claude, not Python. But here's the equivalent
# CLI flow you'd run to add and inspect the server from a terminal:

# 1. Add the remote Scavio MCP (HTTP transport, x-api-key auth)
claude mcp add --transport http scavio https://mcp.scavio.dev/mcp \
  --header "x-api-key: YOUR_API_KEY"

# 2. List configured servers and confirm scavio is there
claude mcp list

# 3. Inside an interactive Claude Code session, verify the tools loaded
#    /mcp  ->  scavio  connected

JavaScript Example

JavaScript

// Same config, JSON form. Drop this into the mcpServers block of your
// Claude Code / Cursor / Windsurf / VS Code MCP settings file.
{
  "mcpServers": {
    "scavio": {
      "url": "https://mcp.scavio.dev/mcp",
      "headers": {
        "x-api-key": "YOUR_API_KEY"
      }
    }
  }
}

Expected Output

JSON

After `/mcp`, Claude Code shows:

  scavio  connected
  tools: search_google, reddit_search, reddit_post, youtube_search, youtube_metadata, amazon_product, walmart_product, tiktok_profile, tiktok_search_videos, tiktok_hashtag, ... (33 tools total)

Then, asking Claude to "search Google for X" returns live organic results, related searches, and (with the deeper mode) people-also-ask and knowledge-graph data — pulled fresh, not from the model's training cutoff.

Set up the Scavio MCP server in Claude Code (5-minute guide)

Prerequisites

Walkthrough

Step 1: Get your Scavio API key

Step 2: Add the server with one command

Step 3: Or edit the JSON config directly

Step 4: Verify with /mcp before you trust it

Step 5: Ask Claude to search

Python Example

JavaScript Example

Expected Output

Related Tutorials

Frequently Asked Questions

How long does this set up the scavio mcp server in claude code (5-minute guide) tutorial take?

What do I need before starting?

Can I run this tutorial with the free tier?

What frameworks does this work with?

Related Resources

Best Search API for Claude Code in 2026

MCP Custom Search Server

IDE MCP Search

Claude Code Web Search via Scavio MCP

Best MCP Search Tools for Claude Code in 2026

Scavio MCP vs Built-in MCP web_search

Start Building

Set up the Scavio MCP server in Claude Code (5-minute guide)

Prerequisites

Walkthrough

Step 1: Get your Scavio API key

Step 2: Add the server with one command

Step 3: Or edit the JSON config directly

Step 4: Verify with /mcp before you trust it

Step 5: Ask Claude to search

Python Example

JavaScript Example

Expected Output

Related Tutorials

Frequently Asked Questions

How long does this set up the scavio mcp server in claude code (5-minute guide) tutorial take?

What do I need before starting?

Can I run this tutorial with the free tier?

What frameworks does this work with?

Related Resources

Best Search API for Claude Code in 2026

MCP Custom Search Server

IDE MCP Search

Claude Code Web Search via Scavio MCP

Best MCP Search Tools for Claude Code in 2026

Scavio MCP vs Built-in MCP web_search

Start Building