◉CLAUDE LABJP
●FORK — Claude Code 2.1.212 changes what /fork does: it copies your conversation into a new background session with its own row in claude agents, so you can keep working. The old in-session subagent is now /subtask●LIMITS — WebSearch calls are now capped at 200 per session by default, and subagent spawns get the same 200 ceiling, so a runaway search or delegation loop stops on its own●MCPBG — MCP tool calls running past two minutes now move to the background automatically, keeping the session usable. Tune the threshold with CLAUDE_CODE_MCP_AUTO_BACKGROUND_MS●PLANFIX — Fixed plan mode auto-running file-modifying Bash commands such as touch and rm without a permission prompt or an SDK canUseTool callback●SONNET5 — Claude Sonnet 5 is running on introductory pricing of $2 per million input tokens and $10 per million output. After August 31 it moves to $3 and $15●IPO — Bankers are reportedly lining up investor meetings for Anthropic ahead of a possible public listing as soon as October●FORK — Claude Code 2.1.212 changes what /fork does: it copies your conversation into a new background session with its own row in claude agents, so you can keep working. The old in-session subagent is now /subtask●LIMITS — WebSearch calls are now capped at 200 per session by default, and subagent spawns get the same 200 ceiling, so a runaway search or delegation loop stops on its own●MCPBG — MCP tool calls running past two minutes now move to the background automatically, keeping the session usable. Tune the threshold with CLAUDE_CODE_MCP_AUTO_BACKGROUND_MS●PLANFIX — Fixed plan mode auto-running file-modifying Bash commands such as touch and rm without a permission prompt or an SDK canUseTool callback●SONNET5 — Claude Sonnet 5 is running on introductory pricing of $2 per million input tokens and $10 per million output. After August 31 it moves to $3 and $15●IPO — Bankers are reportedly lining up investor meetings for Anthropic ahead of a possible public listing as soon as October
← Back to Blog

Claude AI in March 2026 — Monthly Roundup and What to Watch in April

monthly-roundupclaude-updatesauto-modecomputer-use2026

Introduction

March 2026 has been a defining month for Claude AI. Building on the momentum of the Opus 4.6 and Sonnet 4.6 launches in February, Anthropic rolled out a series of features that fundamentally change how we interact with AI — shifting from conversation to action.

In this roundup, we'll cover the month's biggest updates, highlight the most impactful articles published on Claude Lab, and look ahead to what April has in store.

Major Updates This Month

Claude Code Auto Mode

On March 24, Anthropic introduced Auto Mode for Claude Code. Previously, every file edit and command execution required manual approval. Auto Mode changes that by letting the AI evaluate the risk of each action in real time — safe operations proceed automatically, while potentially dangerous ones are blocked.

The feature includes built-in safeguards that detect risks such as unintended behavior and prompt injection attacks. It's currently available as a research preview for Sonnet 4.6 and Opus 4.6, and Anthropic recommends using it in sandboxed environments for now.

Computer Use — Claude Takes the Desktop

Also announced on March 24, Computer Use gives Claude direct keyboard-and-mouse control of your macOS desktop through Cowork. You can message Claude a task from your phone, and it will open apps, navigate browsers, fill in spreadsheets, and more — all on its own.

This feature initially launched for Claude Pro and Claude Max subscribers on macOS. When native connectors are available (for services like Google Workspace and Slack), Claude uses those first for better reliability, falling back to direct desktop control when needed.

1M Token Context Window Goes GA

The 1 million token context window became generally available for Opus 4.6 and Sonnet 4.6 at standard pricing. This means you can feed Claude an entire book or a full codebase in a single request — a capability that opens up entirely new workflows for research, code analysis, and document processing.

128K Output Token Expansion

Opus 4.6 now supports up to 128K output tokens, while Sonnet 4.6 goes up to 64K. Whether you're generating long-form documentation, complete codebases, or detailed analyses, the expanded output makes it practical to get comprehensive results in a single pass.

Anthropic's 81,000-Person AI Survey

Anthropic published the results of a landmark study in which nearly 81,000 Claude users shared their hopes and concerns about AI. It's the largest multilingual qualitative study of its kind, offering valuable insights into what people around the world actually want from artificial intelligence.

Claude Lab by the Numbers

We published 329 articles across four categories this month:

  • Claude AI (108 articles): From beginner guides to advanced strategies, covering the full spectrum of Claude usage
  • Claude Code (103 articles): Deep dives into Auto Mode, multi-agent orchestration, CI/CD integration, and developer workflows
  • API & SDK (62 articles): Everything from Agent SDK quickstarts to production-grade system design patterns
  • Cowork (56 articles): Practical guides for browser automation, Canva integration, AI-powered side businesses, and more

Five Must-Read Articles from March

  1. "Claude Code / API Max Output Tokens Expanded to 128K" — Practical strategies for making the most of the new output limits
  2. "Claude Agent SDK Multi-Agent Design Patterns" — An architectural guide to coordinating multiple AI agents
  3. "Figma MCP × Claude Code" — A complete workflow from design to code generation
  4. "What 81,000 People Told Anthropic About AI" — Key takeaways from the largest AI perception study ever conducted
  5. "AI Side Business Guide — Monetization Strategies with Claude and Multiple AIs" — A fresh perspective on building income streams with AI tools

Reflections on March

Looking back at this month, one theme stands out clearly: Claude is moving from conversation to action. Auto Mode represents a shift from "asking AI for help" to "delegating tasks to AI," and Computer Use takes that a step further into a world where AI operates autonomously on your desktop.

Meanwhile, the Agent SDK has lowered the barrier to building custom multi-agent systems. We're entering an era where the skill isn't just "how to use AI" — it's "how to design teams of AI agents that work together."

These aren't incremental improvements. They signal a fundamental change in the relationship between people and AI tools.

What to Watch in April

Claude Haiku 3 Deprecation (April 19)

Claude Haiku 3 will be officially retired on April 19. If you're still using it in production, now is the time to migrate to Haiku 4.5. We've published a detailed migration guide on Claude Lab to help with the transition.

Auto Mode and Computer Use Maturation

Both features launched as research previews in March. April should bring improved stability, broader platform support, and potentially Windows compatibility for Computer Use — something many users are eagerly waiting for.

Growing Agent Ecosystem

The MCP server ecosystem is expanding rapidly. April is likely to bring deeper integrations with Slack, GitHub, Google Workspace, and other services, making it easier than ever to automate complex workflows with Claude.

Wrapping Up

March 2026 has been a month of transformation for Claude AI. From a powerful chat assistant to a coding partner, desktop operator, and autonomous agent — Claude's role continues to expand in ways that were hard to imagine just a year ago.

We'll keep bringing you the latest news, guides, and insights throughout April. If there's a topic you'd like us to cover or a question you'd like answered, don't hesitate to reach out.

Thank you for reading, and here's to another great month ahead.