tools
A set of tools that gives agents powerful capabilities.
claude mcp add --transport stdio strands-agents-tools python -m mcp_server \ --env MCP_LOG_LEVEL="INFO" \ --env MCP_SERVER_CONFIG="Path or inline config for Tools MCP Server"
How to use
Strands Agents Tools provides a comprehensive set of ready-to-use capabilities that can be dynamically loaded and invoked by MCP-enabled agents. The server exposes a rich toolbox including file operations, shell commands, memory storage, web interactions (HTTP requests, web search and content extraction via Tavily-powered tooling), Python execution with safety features, mathematical calculations, AWS integration, image and video processing, audio output, environment management, journaling, task scheduling, advanced reasoning, swarm coordination, and a dynamic MCP client for connecting to remote MCP servers. Agents can call these tools individually or in parallel (Batch Tool) to compose complex workflows without writing boilerplate code. The tools are designed to work across various data sources and environments, enabling agents to persist memories, query knowledge bases, and perform automated web or system tasks as part of problem solving.
To use the server, start it with the provided command and connect your agent framework to the exposed MCP endpoints. Your agent can invoke tools through the standard MCP interface, specifying the tool name (e.g., file_read, shell, http_request) and the appropriate parameters. For example, a workflow could read a configuration file, perform a shell command, fetch data from an API, and then store results in a memory store for subsequent runs. The Tools package also supports dynamic loading of tools from remote MCP servers, allowing agents to leverage externally hosted capabilities with proper security considerations.
How to install
Prerequisites:
- Python 3.9 or newer (recommended 3.10+)
- Access to a Python environment (virtualenv optional but recommended)
Installation steps:
-
Create and activate a virtual environment (optional but recommended): python3 -m venv .venv source .venv/bin/activate # Linux/macOS .venv\Scripts\activate # Windows
-
Install the Tools package (this MCP server component): pip install strands-agents-tools
-
Install development or optional tooling (if you plan to contribute or enable extra features): pip install strands-agents-tools[mem0_memory, use_browser, rss, use_computer]
-
Run the MCP server (adjust as needed for your environment): python -m mcp_server
-
If you are using a custom configuration file, ensure the server can load it (e.g., via environment variable or command-line option) and restart the server accordingly.
Notes:
- Ensure network access between the agent framework and the MCP server.
- Review any security considerations when enabling dynamic MCP client features.
Additional notes
Tips and common issues:
- If you enable the Dynamic MCP Client, be mindful of security: only connect to trusted MCP endpoints and verify tool signatures when possible.
- For long-running tool calls (e.g., large file operations or web crawling), consider batching calls or using the Batch Tool to run parallel operations safely.
- Set environment variables (MCP_LOG_LEVEL, MCP_SERVER_CONFIG) to tune logging and configuration without changing code.
- If you encounter import errors, ensure you are using a compatible Python version and that strands-agents-tools is installed in the active environment.
- Consult the Tools Overview in the README for usage patterns and examples of invoking tools via agents.
Related MCP Servers
bytebot
Bytebot is a self-hosted AI desktop agent that automates computer tasks through natural language commands, operating within a containerized Linux desktop environment.
sdk-python
A model-driven approach to building AI agents in just a few lines of code.
ag2
AG2 (formerly AutoGen): The Open-Source AgentOS. Join us at: https://discord.gg/sNGSwQME3x
koog
Koog is the official Kotlin framework for building predictable, fault-tolerant and enterprise-ready AI agents across all platforms – from backend services to Android and iOS, JVM, and even in-browser environments. Koog is based on our AI products expertise and provides proven solutions for complex LLM and AI problems
samples
Agent samples built using the Strands Agents SDK.
sdk-typescript
A model-driven approach to building AI agents in just a few lines of code.