Strands Agents Moss Integration

Moss delivers sub-10ms semantic retrieval, giving your Strands Agents instant access to a knowledge base during conversations.

Installation

pip install strands-agents-moss

Prerequisites

• Moss project ID and project key (get them from Moss Portal)
• Python 3.10+
• Model provider credentials — Strands Agents defaults to Amazon Bedrock as the LLM provider. Make sure your AWS credentials are configured (e.g. AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_REGION). To use a different provider, see Choosing a model provider below.

Quick Start

import asyncio
import os

from strands import Agent
from strands_agents_moss import MossSearchTool

async def main():
    # Create and pre-load the Moss search tool
    moss = MossSearchTool(
        project_id=os.getenv("MOSS_PROJECT_ID"),
        project_key=os.getenv("MOSS_PROJECT_KEY"),
        index_name="my-index",
    )
    await moss.load_index()

    # Create a Strands agent with Moss retrieval
    agent = Agent(tools=[moss.tool])
    agent("What is your refund policy?")

asyncio.run(main())

Choosing a Model Provider

Strands Agents defaults to Amazon Bedrock. If you don't have AWS credentials or prefer a different provider, pass a model argument to Agent:

# OpenAI
from strands.models.openai import OpenAIModel
agent = Agent(model=OpenAIModel("gpt-4o"), tools=[moss.tool])

# Anthropic
from strands.models.anthropic import AnthropicModel
agent = Agent(model=AnthropicModel("claude-sonnet-4-20250514"), tools=[moss.tool])

See the Strands model providers docs for all supported providers.

Configuration Options

MossSearchTool

Methods

Multi-Agent Example

Moss tools work seamlessly with Strands' agents-as-tools pattern:

from strands import Agent
from strands_agents_moss import MossSearchTool

async def main():
    moss = MossSearchTool(
        index_name="product-docs",
    )
    await moss.load_index()

    # Research agent with knowledge base access
    researcher = Agent(
        system_prompt="You are a research assistant. Use moss_search to find information.",
        tools=[moss.tool],
    )

    # Orchestrator that delegates to the researcher
    orchestrator = Agent(
        system_prompt="You coordinate research tasks. Delegate questions to the researcher.",
        tools=[researcher.as_tool(
            name="researcher",
            description="A research assistant with access to the knowledge base",
        )],
    )

    orchestrator("Summarize our return and refund policies.")

License

This integration is provided under the BSD 2-Clause License.

Support

• Moss Docs
• Moss Discord
• Strands Agents Docs