Google ADK Integration
Integrate Scavio with the Agent Development Kit (ADK) to give any Google ADK agent real-time search across Google, YouTube, Amazon, Walmart, and Reddit -- via the native google-adk-scavio package, or the hosted Scavio MCP server for the full 46-tool catalog including TikTok and Instagram.
Native ADK tools
google-adk-scavio registers plain typed Python functions that ADK wraps automatically -- no MCP session, no Node.js, just pip install.Option A: Native package (recommended)
Step 1: Install
pip install google-adk-scavioStep 2: Get your API key
Grab a key at dashboard.scavio.dev (50 free signup credits, no credit card) and export it as SCAVIO_API_KEY.
Step 3: Add the tools to an agent
from google.adk.agents import LlmAgent
from google_adk_scavio import create_scavio_tools
agent = LlmAgent(
model="gemini-2.5-flash",
name="search_agent",
instruction="Use Scavio to answer questions with fresh web data.",
tools=create_scavio_tools(), # reads SCAVIO_API_KEY
)The factory registers 16 tools (Google search and news, YouTube, Amazon, Walmart, Reddit, plus scavio_extract for any URL) named scavio_<provider>_<action>. Disable providers you do not need: create_scavio_tools(enable_reddit=False).
Map, crawl and batch-extract websites
Pass enable_crawl=True to add five website tools: scavio_map_website (a site's page URLs from its sitemaps and start page), scavio_crawl_website and scavio_crawl_status (crawl a site from one URL), and scavio_batch_extract and scavio_batch_status (extract a list of URLs in one job). Crawl and batch wait up to wait_seconds (default 45) and return the job id with the first results; the status tools read the rest for free. Map is 1 credit in normal mode; crawl and batch bill per successful page (1, 1 or 2 credits by mode). Requires google-adk-scavio 0.4 or later. API reference: map, crawl, crawl status, batch and batch status; plan limits are on the crawl and batch pages, and single-page reads use Extract.
tools = create_scavio_tools(enable_crawl=True)Option B: Hosted MCP server (full catalog)
ADK connects to external tools over the Model Context Protocol. Point an McpToolset at mcp.scavio.dev and your agent gets all 46 Scavio tools, including TikTok and Instagram. Requires pip install "google-adk[mcp]". It is a cost-effective Tavily and SerpAPI alternative.
Connect the Scavio MCP server
from google.adk.agents import LlmAgent
from google.adk.tools.mcp_tool.mcp_toolset import McpToolset
from google.adk.tools.mcp_tool.mcp_session_manager import (
StreamableHTTPConnectionParams,
)
agent = LlmAgent(
model="gemini-2.0-flash",
name="search_agent",
instruction="Use Scavio to answer questions with fresh web data.",
tools=[
McpToolset(
connection_params=StreamableHTTPConnectionParams(
url="https://mcp.scavio.dev/mcp",
headers={"x-api-key": "YOUR_SCAVIO_API_KEY"},
),
)
],
)Run locally instead (stdio)
Prefer to run the MCP server on your own machine? Use StdioConnectionParams with the @scavio/mcp-server npm package (Node.js 20+):
from google.adk.tools.mcp_tool.mcp_session_manager import (
StdioConnectionParams,
StdioServerParameters,
)
toolset = McpToolset(
connection_params=StdioConnectionParams(
server_params=StdioServerParameters(
command="npx",
args=["-y", "@scavio/mcp-server"],
env={"SCAVIO_API_KEY": "YOUR_SCAVIO_API_KEY"},
)
)
)Available Tools
Once connected, your agent can call the full Scavio catalog -- Google Search, News, Maps, Shopping, YouTube, Amazon, Walmart, Reddit, TikTok, and Instagram. See the MCP reference for the complete list.
Benefits of Scavio + Google ADK
- Zero-install: connect a hosted MCP URL.
- Full catalog: every Scavio tool, auto-converted.
- Local or remote: stdio or Streamable HTTP.
- Cost-effective: most calls cost a single credit.
Next Steps
- MCP Integration -- all clients and the full catalog
- Google Search API -- endpoint reference