Search Tools

This guide shows how to add RedisVL-backed search tools to a Google ADK agent.

Prerequisites

• Redis 8.4+ with the RediSearch module, or Redis Cloud with Search.
• adk-redis with the search extra: pip install 'adk-redis[search]'.
• A RedisVL index with the documents you want to search.

In-process Python tools

All five tools subclass BaseRedisSearchTool and accept a bound RedisVL SearchIndex / AsyncSearchIndex.

Vector search

from google.adk import Agent
from redisvl.index import SearchIndex
from redisvl.utils.vectorize import HFTextVectorizer

from adk_redis import RedisVectorQueryConfig, RedisVectorSearchTool

vectorizer = HFTextVectorizer(model="sentence-transformers/all-MiniLM-L6-v2")
index = SearchIndex.from_existing("products", redis_url="redis://localhost:6379")

tool = RedisVectorSearchTool(
    index=index,
    vectorizer=vectorizer,
    config=RedisVectorQueryConfig(
        vector_field_name="embedding",
        num_results=5,
    ),
    return_fields=["title", "price", "category"],
    name="search_products",
    description="Semantic search across the product catalog.",
)

agent = Agent(model="gemini-2.0-flash", tools=[tool])

Hybrid search

Combines vector similarity with BM25 keyword scoring. On Redis 8.4+ with redisvl>=0.18.2, the tool uses the native FT.HYBRID command. On older versions or older Redis it falls back to client-side aggregation.

from adk_redis import RedisHybridQueryConfig, RedisHybridSearchTool

tool = RedisHybridSearchTool(
    index=index,
    vectorizer=vectorizer,
    config=RedisHybridQueryConfig(
        text_field_name="description",
        combination_method="LINEAR",
        linear_alpha=0.7,  # 70% text, 30% vector
        num_results=10,
        stopwords=None,  # set to a language string when nltk is installed
    ),
)

Range search

Returns every document within a distance threshold of the query vector.

from adk_redis import RedisRangeQueryConfig, RedisRangeSearchTool

tool = RedisRangeSearchTool(
    index=index,
    vectorizer=vectorizer,
    config=RedisRangeQueryConfig(
        vector_field_name="embedding",
        distance_threshold=0.3,
        num_results=20,
    ),
)

Text search

BM25-only keyword search. No vectorizer required.

from adk_redis import RedisTextQueryConfig, RedisTextSearchTool

tool = RedisTextSearchTool(
    index=index,
    config=RedisTextQueryConfig(text_field_name="content"),
    return_fields=["title", "content"],
)

SQL search

RedisSQLSearchTool wraps redisvl.query.SQLQuery. The LLM passes a SQL SELECT statement and an optional params object; the tool translates it into the right FT.SEARCH or FT.AGGREGATE call.

from adk_redis import RedisSQLSearchTool

tool = RedisSQLSearchTool(
    index=index,
    sql_redis_options={"schema_cache_strategy": "lazy"},
)

Example SQL the agent might emit:

SELECT title, price, category
FROM products
WHERE category = 'electronics' AND price < :max_price

with params={"max_price": 50}. Install with pip install 'adk-redis[sql]'.

RedisVL MCP server

Connect an ADK agent to RedisVL's own MCP server (rvl mcp) using ADK's standard McpToolset. The server exposes two tools per index:

• search-records: schema-aware search (vector / fulltext / hybrid, chosen at server start). Filter and return-field hints come from the bound index schema.
• upsert-records: write path. Suppress with --read-only on the server, or with tool_filter=["search-records"] on the toolset.

from google.adk import Agent
from google.adk.tools.mcp_tool import McpToolset
from google.adk.tools.mcp_tool.mcp_session_manager import StdioConnectionParams
from mcp import StdioServerParameters

# In-process stdio: spawn `rvl mcp --config <path>` next to the agent.
agent = Agent(
    model="gemini-2.5-flash",
    name="redis_mcp_agent",
    tools=[
        McpToolset(
            connection_params=StdioConnectionParams(
                server_params=StdioServerParameters(
                    command="rvl",
                    args=[
                        "mcp",
                        "--config",
                        "/etc/redisvl/mcp.yaml",
                        "--read-only",
                    ],
                ),
                timeout=30,
            ),
            tool_filter=["search-records"],
        ),
    ],
)

For an already-running remote server, swap StdioConnectionParams for StreamableHTTPConnectionParams(url=..., headers={"Authorization": "Bearer ..."}) (or SseConnectionParams).

Install the MCP CLI with pip install 'redisvl[mcp]>=0.18.2' and start the server with rvl mcp --config <path> --transport streamable-http (or stdio).

Note

For other ADK language SDKs (TypeScript, etc.), see Custom MCP Tools.

When to use which

Path	Use when
In-process tools (RedisVectorSearchTool, etc.)	Single Python agent, fine-grained control, no extra service to operate.
McpToolset against rvl mcp	Multiple agents (Python, JS, Claude Desktop) share one index, schema-aware tool descriptions, you want --read-only as a deployment-level guardrail.

Notes

• All in-process search tools accept custom name and description so the LLM sees a domain-specific tool.
• epsilon is a VECTOR_RANGE-only attribute, so RedisVectorQueryConfig does not accept it. Use RedisRangeQueryConfig.epsilon for range searches.
• The hybrid tool auto-detects native FT.HYBRID support. Pass a RedisAggregatedHybridQueryConfig explicitly to force the older aggregation path.