Claude Code Ships an Accessibility-First Release
The latest Claude Code update leads with inclusion. There’s a new opt-in screen reader mode that swaps the rich terminal UI for plain-text rendering friendlier to assistive tech, plus vim insert-mode remaps, corporate proxy process wrapping, and mouse-click support inside fullscreen menus. Small on their own, together they widen who can actually drive an agentic CLI comfortably.
Under the hood, the release keeps grinding on reliability: improved background agents, cleaner CLI output, better usage tracking, and a batch of stability, performance, and memory optimizations. A notable fix unblocks /model and other dialogs that were getting stuck in background agent sessions — a real annoyance for anyone running long autonomous runs.
Claude Science AI for Science Grants Close Tomorrow
Two weeks after launching Claude Science, its AI workbench for researchers, Anthropic’s AI for Science compute-grant program is at the wire. The program will fund up to 50 projects with up to $30,000 in Claude credits each, with Modal chipping in up to $2,000 of compute for select projects. Graduate and postdoc researchers with active work can apply, and Anthropic is steering toward biology and biomedical research.
The clock matters: applications close July 15, award notifications go out by July 31, and funded projects run September 1 through December 1. If you’ve been circling an application, this is your last full day.
SDKs Add Code Execution With REPL State Persistence
The Python, TypeScript, Go, Java, Ruby, PHP, and C# SDKs now support code_execution_20260120, the tool version that adds REPL state persistence — variables and imports survive across calls instead of resetting each time. It’s also the minimum version required for programmatic tool calling, so it’s not just a nicety if you’re building agent loops.
Adoption is refreshingly low-friction: set the tool’s type to code_execution_20260120 and go — no beta header required. Persistent state makes multi-step data work and iterative analysis far more natural for tool-using agents.
LTM and UST Deepen the Claude Enterprise Push
The Claude Partner Network keeps filling out. LTM announced a partnership to accelerate enterprise-scale adoption of Claude, Claude Code, and Claude Cowork across engineering, modernization, and business workflows, folding the tools into its implementation practice to help clients move from pilots to production. It follows UST’s alliance days earlier to embed Claude into engineering environments for Global 1000 firms.
The context: Anthropic has committed an initial $100 million to the Claude Partner Network for 2026, and since launch in March, more than 40,000 firms have applied and over 10,000 consultants have earned a Claude certification. The distribution flywheel for enterprise Claude is spinning up fast.
Reflect Is a Retention Play Dressed as a Wellbeing Feature
Anthropic’s new Claude Reflect dashboard — a “Claude Wrapped” that surfaces your top topics, peak hours, and AI habits, then nudges you to set quiet hours and take breaks — is getting a skeptical read, and fairly so. As TechCrunch put it, a feature that shows you how much you rely on Claude, at a moment of rising AI backlash, is also quietly selling you on AI. The more your workflow lives inside the dashboard, the harder it is to leave.
That doesn’t make it cynical — it makes it clever. The wellbeing framing is real (quiet hours, reflection prompts like “what’s one thing you want to keep doing yourself?”), and it’s also perfectly aligned with retention. It’s the same pattern as this month’s bigger moves: Cowork in the cloud, write access to your inbox, Reflect in your habits. Each one makes Claude a little more load-bearing in the day. The honest version of the pitch is that mindful use and daily dependence can be the same product — and Anthropic is betting they are.