strands-agents/tools

Strands Agents Tools – a toolbox for AI agents

What it is – a Python package that ships a large collection of ready‑made tools an LLM‑based agent can call. The tools let an agent read/write files, run shell commands, query web search services, call APIs, interact with Slack, execute Python code, perform math, generate images/video/audio, manage AWS resources, store/retrieve memories, schedule jobs, drive a browser or even control the desktop. In short, it bridges the gap between a language model and real‑world actions.

Why it matters – modern AI agents need a way to act on the outside world. Strands Agents Tools provides a model‑driven interface: you declare the tools you want an agent to have, and the agent can invoke them directly from its generated responses. This lets developers build sophisticated workflows (e.g., “search the web → scrape a page → run a Python analysis → store the result in a vector store”) with only a few lines of code.

---

Core features (as listed in the README)

Category	Tools (selected)	Typical use case
File I/O	file_read, file_write, editor	Load config files, write results, batch‑edit code snippets
Shell / System	shell, cron	Run OS commands, schedule recurring jobs (Linux only)
Web & Search	tavily_search, exa_search, browser	Real‑time web search, site crawling, automated browsing
HTTP / API	http_request	Call REST endpoints, fetch data from services
Messaging	slack	Post messages or react to Slack events
Python execution	python_repl, code_interpreter	Execute arbitrary Python (with safety prompts) or sandboxed multi‑language code
Math	calculator	Symbolic or numeric calculations
AWS	use_aws	Interact with S3, DynamoDB, etc.
Image / Video / Audio	generate_image, generate_image_stability, nova_reels, speak	Create AI‑generated media or play back audio
Memory	mem0_memory, mongodb_memory, elasticsearch_memory, agent_core_memory	Persist user/agent context across runs
Advanced reasoning	think, agent_graph, graph, workflow	Build multi‑step, multi‑agent pipelines with deterministic DAGs
Swarm / Multi‑agent	swarm, use_agent, use_llm	Coordinate several agents, nest agents with model switching
Desktop automation	use_computer, browser	Move mouse, click, type, capture screenshots
Utility	environment, journal, current_time, stop, handoff_to_user	Manage env vars, logging, graceful termination

Security note – many tools give agents powerful capabilities (file system access, code execution, AWS calls, remote MCP connections, browser automation). The README stresses that these are experimental and should be reviewed for safety before production use.

---

Quick start

# Install the core package
pip install strands-agents-tools

# Install optional extras you need (e.g., browser automation, RSS, memory back‑ends)
pip install "strands-agents-tools[mem0_memory,use_browser,rss,use_computer]"

Minimal usage example (from the README)

from strands import Agent
from strands.tools import file_read, python_repl, use_aws

# Create an agent with the desired tools
agent = Agent(tools=[file_read, python_repl, use_aws])

# Agent can now call the tools directly in its prompts
response = agent.run(
    "Read the config file at /etc/app/config.yaml, extract the API key, then list all S3 buckets using that key."
)
print(response)

The Agent class (provided by the Strands SDK) automatically maps the natural‑language request to the appropriate tool calls.

---

When to use it

• Prototyping AI assistants that need to read/write data, call external services, or control a computer.
• Building multi‑agent pipelines where each step is a distinct tool (e.g., search → scrape → analyze → store).
• Experimenting with advanced reasoning by chaining tools via the built‑in DAG/graph utilities.
• Integrating with existing cloud stacks (AWS, MongoDB, Elasticsearch) through the provided memory and service tools.

---

Where to go next

• Check the official documentation at https://strandsagents.com/ for detailed tool signatures and safety guidelines.
• Explore the samples repository (github.com/strands-agents/samples) for ready‑made agent scripts.
• Join the Strands Discord (link in the badge) for community support and updates.