Claude Code Launches Agent View for Parallel Session Management
Anthropic shipped Agent View—a unified CLI dashboard for managing parallel Claude Code sessions—and revealed doubled rate limits for all paid plans at its annual Code w/ Claude SF developer conference.
Image by CWA
Claude Code Launches Agent View for Parallel Session Management—and Doubles Rate Limits
Anthropid used this week's Code w/ Claude SF developer conference to ship two updates that address long-standing friction points for Claude Code power users: Agent View, a unified CLI dashboard for managing parallel sessions, and doubled rate limits across all paid plan tiers.
Both changes are live now.
Agent View: One Place to Manage Every Claude Code Session
Shipped on May 11, Agent View is the clearest signal yet that Anthropic sees multi-agent orchestration as a first-class workflow, not an edge case.
Before Agent View, running parallel Claude Code sessions meant juggling multiple terminal tabs, a tmux grid, and an informal mental ledger of which agents needed attention and which were still working. There was no single view. If Claude was waiting on a decision in a background session, you either caught it or you didn't.
Agent View changes that structure at the CLI level. Press the left arrow from any session—or run claude agents from the terminal—to open a dashboard where every active session shows up as a row. Each row surfaces the session status, the last response, and how long ago you last interacted with it. Sessions that need your input flag themselves; sessions that don't stay parked in the background until they do.
Replying to a waiting session no longer requires switching context. Select the session, answer inline, and it picks back up without forcing you out of what you were doing. For sessions where you want to read the full transcript, press Enter to attach directly.
Anything can be backgrounded. Use /bg inside an active session to push it to the background, or skip the foreground entirely with claude --bg [task] to kick off a fresh background session from the command line.
Agent View is available as a Research Preview on Pro, Max, Team, Enterprise, and API plans. Standard rate limits apply.
Doubled Rate Limits Across All Paid Tiers
Announced at the Code w/ Claude SF conference, Anthropic has doubled the five-hour rate limits for Claude Code on Pro, Max, Team, and Enterprise plans. API rate limits for Claude Opus were also raised.
The capacity increase is a direct result of Anthropic's compute deal with SpaceX, which put over 220,000 NVIDIA GPUs at the Colossus 1 data center online for Anthropic's use. Along with the rate limit increase, Anthropic removed the peak-hour limit reductions that had been applied to Pro and Max users during high-demand periods.
For developers running long-horizon coding workflows—multi-step refactors, test generation suites, background agents monitoring PRs—hitting a rate ceiling mid-session has been one of the more disruptive interruptions in an otherwise productive flow. The doubled limits directly address that.
What Happened at Code w/ Claude SF
Code w/ Claude SF 2026 ran on May 12 in San Francisco, with two days of keynotes, breakout sessions, and workshops. All Day 1 sessions are now on YouTube.
The conference featured customer sessions from Asana, Cursor, GitHub, Replit, and Vercel on how their teams are designing production-grade agents, alongside Anthropic team talks on model selection, skill architecture, and scaling AI-native engineering organizations.
The conference continues in London (May 19-20) and Tokyo (June 5-6), with Day 1 keynotes and breakout sessions streaming live.
The Bigger Picture
Agent View and the doubled rate limits are part of a broader push from Anthropic this week. Other recent Claude Code and platform updates include the Claude Platform on AWS GA (May 11), Claude Managed Agents' new Dreaming and multiagent orchestration features (May 6), and the Claude for Microsoft 365 launch across Excel, PowerPoint, Word, and Outlook.
The common thread is moving Claude Code from a single-session pair-programming tool toward something closer to an orchestration layer: parallel agents, background tasks, memory that persists across runs, and infrastructure that can scale with the teams using it. Agent View gives that pattern a first-class CLI interface. The doubled limits give it the headroom to actually run.
For developers who have already been pushing Claude Code to its limits, both changes land as quality-of-life upgrades to workflows that were already happening. For teams evaluating Claude Code for production use, they reduce two of the most frequently cited concerns: operational complexity and session caps.
