The Pulse

Strands Harness SDK Adds Built-In File, Web and Multi-Agent Tools for Python and TypeScript

Strands says its Harness SDK now includes built-in tools for file operations, web research and multi-agent workflows in Python and TypeScript. The update also deprecates the separate Python tools package, with legacy imports set to raise an

AI.info Team · Written with AI · fact-checked · 2026-11-06

Strands says it has expanded the built-in tools available in its Harness SDK for agents written in Python and TypeScript. The tools cover file operations, command execution, web-page retrieval and delegation to other agents. Developers can add these capabilities to custom agents without building and maintaining each tool themselves.

The update addresses tasks Strands identifies as common across different kinds of agents: reading files, fetching web content, running commands and delegating work. Its examples include customer support, incident triage and deep research agents. The tools are available in both language versions of the SDK.

File tools and shell

The SDK includes tools for reading, writing and editing files. Strands describes pairing them with the shell tool when an agent needs to run commands, write and execute code, or process data. The file tools cover working with the contents of files; the shell tool supports command-based tasks.

For incident triage, Strands gives the example of an agent inspecting exported logs, running diagnostic commands and saving an investigation summary.

Web Fetch and Subagent

Web Fetch is intended for research agents gathering information from web pages. It converts page content into markdown that a model can read. Strands also describes using Subagent to delegate part of a research task to a child agent, provide that agent with the tools it needs and receive a summary in return.

The intermediate research remains in the child agent’s own context, according to Strands. In the described workflow, the main agent can assign part of the research rather than conduct every step itself, then use the returned summary.

Graph and Swarm workflows

Strands presents Graph and Swarm as tools for different multi-agent workflows. A document-review agent could use Graph to run factual and editorial reviews in parallel, then pass the feedback to an editor. That example combines two review tasks before the editing step.

Swarm supports a more iterative process. In Strands’ example, an editor can hand a draft back to a writer, or bring in a researcher when a claim needs evidence. Developers can configure each specialist’s role and tool access, so the agents involved in a workflow can be set up for their assigned tasks.

Deprecation and migration

With the tools supported directly in the SDK, Strands is deprecating the strands-agents-tools Python package and its repository. Versions before v0.9.0 will continue working, but will no longer receive updates after November 4, 2026. In v0.9.0, legacy tool imports will raise an error.

Strands tells developers to check the migration guide before upgrading because some replacements also change behavior. The project says developers can open an issue to request support for additional tools, or join its Discord for questions and discussion.