AI UPDATES THAT
ACTUALLY MATTER

Sonnet 5 is now the default Claude Code model with native 1M-token context. Fork-mode subagents are on by default in interactive sessions — a fork subagent inherits the full conversation and prompt cache with no setup required, toggle off with CLAUDE_CODE_FORK_SUBAGENT=0. v2.1.198 shipped two skill-related changes: the /dataviz skill for chart and dashboard design (includes a runnable color-palette validator), and background agents in worktrees now automatically commit, push the branch, and open a draft PR when they finish — no prompt, no pause. Cross-session @ mentions let you ping another Claude session by name mid-conversation.

Sonnet 5 at 1M-token context changes the unit economics of large-codebase work — a full monorepo fits in context without chunking. Fork subagents on by default means parallel agent work stops requiring explicit configuration for the common case. The worktree auto-PR behavior is the most impactful workflow change for teams: background agents now hand off finished work to human review in GitHub without a second command. The /dataviz skill plugs a real gap — chart design has been a persistent weak point for coding agents, and the color-palette validator makes design iteration testable rather than purely visual.

To use fork subagents: nothing to configure — they're on by default. To disable: set CLAUDE_CODE_FORK_SUBAGENT=0. For the /dataviz skill: type /dataviz in any session to activate; it guides chart type selection, color palette, and accessibility. For worktree auto-PR: run `claude agents` in a worktree-based project; the background agent commits and pushes when done and opens a draft PR — check your GitHub notifications. For cross-session mentions: type @ and a session name in your prompt; Claude routes the message via SendMessage.

Agent Plugins 1.0.0 was published on August 6, 2026 by a Technical Steering Committee comprising AWS, Anysphere (Cursor), Microsoft, OpenAI, and Vercel. The spec lives at github.com/agentplugins/agent-plugins-spec. Codex CLI v0.147.0 shipped August 7 as the first implementation — it installs portable Agent Plugins, searches local/personal/workspace/remote plugin catalogs, and imports Cursor-managed skills. Launch clients: ChatGPT, Codex, Cursor, GitHub Copilot, Kiro, and VS Code.

Plugins do not replace skills. A plugin's `skills/` directory holds standalone skills unchanged — the manifest adds an `mcp.json` alongside it so clients discover the bundled MCP server automatically. The practical effect: skills you install today are forward-compatible with the Agent Plugins standard. The spec is currently labeled 'Working Draft' — no permission model, no sandboxing, and no signature verification are defined yet. Those are listed as future work. Ship Claude skills now; they will be valid Agent Plugins when the standard finalises.

To install a plugin via Codex CLI v0.147.0: `codex plugin install <owner>/<repo>`. For Claude Code: `claude plugin marketplace add <owner>/<repo>` still works for skills, and plugins that include an Agent Plugin manifest are handled automatically. Browse plugins at github.com/agentplugins/agent-plugins-spec to track compatible packages as they publish. The steering committee forbids any single vendor from holding a majority of seats, making this a genuinely neutral standard rather than a vendor-controlled one.

Claude voice mode now runs on Opus, Sonnet, and Haiku models. Voice can now reach connected tools like Gmail and Slack, and supports significantly more languages. Available across mobile, desktop, and web. Users can switch models mid-conversation.

Voice mode moves from novelty to a real workflow layer. Anthropic explicitly positioned it for: practicing pitch meetings, deciding between offers, reviewing your own process out loud, brainstorming. Combined with connected tool actions, this is Claude Cowork extending into voice-first work.

Open Claude on mobile, desktop, or web. Toggle voice mode. Select model in-conversation. Ensure Gmail/Slack connectors are enabled in Settings → Integrations.

Anthropic and Cognizant expanded their partnership on July 27, 2026, extending Claude deployment to Cognizant's enterprise client base across industries. Follows the earlier PwC (300,000-professional) and TCS partnerships from May/June.

This is Anthropic's 3rd Big Consulting alliance in 90 days (PwC → TCS → Cognizant). Pattern: Anthropic building the implementation partner ecosystem ahead of IPO, mirroring what Salesforce did in 2010s with Deloitte/Accenture. If your organization uses Cognizant for tech services, expect Claude-based delivery over the next 6 months.

If you work with Cognizant as an enterprise client, ask your account team about Claude-integrated delivery.