Anthropic makes Claude's agent APIs generally available

Anthropic has made computer use, a new browser-use tool, the Skills API, and the Files API generally available on the Claude Platform. The release turns several agent primitives into a production-oriented stack: an agent can operate software without a dedicated integration, load versioned team procedures, and read or return persistent files.

Claude's computer use now works across software and the web

The updated computer-use tool lets Claude take several actions per turn instead of stopping after one model call. It can work from screenshots by clicking, typing, and scrolling in applications that expose no API. Anthropic also adds a browser-use tool that combines visual context with page structure, allowing an agent to target a field or button rather than depend only on screen coordinates.

That distinction matters for workflows built around legacy portals and internal tools. A claims workflow, for example, can read an intake document, navigate an insurer's web application, and save the resulting confirmation. Anthropic says one customer cut its longest claims workflow from 32 minutes to 13 and reported 100% completion in its tested workflows; those figures are vendor-reported, not an independent benchmark.

Skills API and Files API fill the reusable-workflow gap

The Skills API lets developers upload and version folders containing instructions, scripts, and templates, then attach a selected skill to a request. Claude loads the skill when the task needs it and runs it in the code-execution sandbox. This gives teams a way to package operating procedures without hosting a separate workflow service for every specialization.

The Files API provides storage for documents an agent reads and writes. Developers can upload a PDF or spreadsheet once, reference it by ID in later requests, and download generated files. The general-availability release adds automatic file expiration, raises rate limits fivefold, and provides 1 TB of storage per organization.

Availability, compliance, and migration details

The four capabilities are available on the Claude Platform now. Anthropic says the Skills API and Files API also work through Microsoft Foundry. The updated computer-use and browser-use tools are coming soon to Google Cloud's Vertex AI, while existing beta integrations continue to work during migration.

Computer use is also eligible for HIPAA-regulated workloads under Anthropic's business associate agreement. That does not remove the need for application-level controls: browser agents can still submit forms, expose data, or take irreversible actions, so production deployments should keep permissions narrow and require confirmation for consequential steps.

Evidence

Anthropic's announcement establishes the availability date and feature scope. Independent coverage from Kylon describes the same four-part general-availability release and its practical impact on production agents. Anthropic's documentation provides the implementation surface for Skills, including custom skill uploads and the requirement to use the feature with code execution.

For teams already using the beta tools, the immediate work is to check endpoint names, storage-retention behavior, rate limits, and cloud-provider availability before changing production routing. The main shift is availability: these agent building blocks are now intended for production integration on Claude Platform, while some hosted-provider variants remain on a separate rollout path.