Code with Claude 2026: Anthropic's Developer Conference Is Happening Now

If you've been waiting for a signal that AI-assisted software development has gone mainstream, here it is: Anthropic just launched its second annual Code with Claude developer conference — and the first stop is San Francisco, today, May 6, 2026.

This isn't a marketing keynote. It's a full-day, hands-on technical event built for developers who want to go from "using Claude as a chat tool" to building and deploying real agentic systems. Whether you're attending in person or catching the livestream, here's everything you need to know — what's being taught, what Claude Opus 4.7 brings to the table, and why this moment matters for anyone learning AI development.

What Is Code with Claude?

Code with Claude is Anthropic's developer conference series — a direct response to the enormous demand from engineers who want deeper, practical guidance on building with Claude. The format is structured, not a trade show floor: each city event is a full day of technical workshops, live capability demos, and one-on-one office hours with Anthropic engineers.

The 2026 Conference Cities and Dates

Demand for the in-person SF event exceeded expectations so significantly that Anthropic added a second day on May 7, specifically for independent developers and early-stage founders. A full virtual livestream is available for all sessions globally, with recordings published after each event.

Key speakers confirmed for the series include Ami Vora (Head of Product at Anthropic), Boris Cherny (Head of Claude Code), and Angela Jiang (Product Lead for the Claude API and SDKs).

What's Actually Being Taught

This is where it gets practical. The conference isn't structured around broad AI philosophy — it's focused on the specific technical skills developers need to build production-grade agentic systems. Three core areas dominate the agenda:

1. Agentic Coding with Claude Code

The biggest theme across all sessions is the shift from AI-as-assistant to AI-as-autonomous-agent. Workshops cover Claude Code (CLI) and Claude Cowork to automate complex debugging, refactoring, and repository navigation — tasks that previously required constant supervision.

What this looks like in practice: you configure Claude Code with a CLAUDE.md file, set granular permissions, add custom skills, and then hand off multi-step engineering tasks. The sessions teach participants how to build a real project layer by layer — from initial configuration to a working production workflow.

This is the core skill that separates developers who "use AI" from developers who deploy AI to do work on their behalf. If you're studying for the Claude Certified Architect (CCA) exam, agentic system design is a primary exam domain — conference sessions map directly to exam objectives.

2. Model Context Protocol (MCP)

MCP has gone from an interesting Anthropic proposal to the industry-standard protocol for connecting AI to external tools and data. Conference workshops walk through:

• Building a custom MCP server from scratch
• Connecting Claude securely to local and remote data sources
• Debugging MCP connections and handling authentication
• Production patterns for MCP in enterprise environments

If you haven't dug into MCP yet, the conference represents the clearest on-ramp available. Anthropic engineers are teaching live implementations — not slides, but actual running code. Check our guide to building MCP servers with Claude for a solid foundation before attending.

3. CLAUDE.md Configuration and Project Architecture

One of the most underrated sessions on the agenda: the deep dive into CLAUDE.md configuration. This is the file that tells Claude Code who it's working with, what tools are available, what rules to follow, and how to behave across a project.

The workshop covers configuring CLAUDE.md at the project level, layering permissions, integrating custom skills, and wiring everything together with a working MCP server. This is production-grade Claude Code engineering — not the "add a comment to this function" use case most people start with.

What Claude Opus 4.7 Changes for Developers

The timing of the conference isn't accidental. Anthropic launched Claude Opus 4.7 on April 16, 2026 — its most capable model to date — and the conference is the first major opportunity to teach developers how to leverage it. Several Opus 4.7 features are central to what's being demonstrated:

High-Resolution Vision (2576px / 3.75MP)

Opus 4.7 is the first Claude model with genuine high-resolution image support, raising the maximum input resolution from 1568px to 2576px (a jump from 1.15MP to 3.75MP). This matters for developers building:

• Code review tools that analyze screenshots and diagrams
• Design-to-code pipelines processing high-fidelity mockups
• Document intelligence systems handling dense visual content
• Medical or scientific imaging workflows

Effort Levels and Task Budgets

Two new developer controls ship with Opus 4.7:

Effort levels let you tune the intelligence/token-spend tradeoff. A new xhigh effort level is the recommended setting for coding and agentic use cases — it produces substantially better results than the default high level on complex, multi-step tasks. Task budgets give Claude a target token count for an entire agentic loop, including thinking, tool calls, tool results, and final output. This lets you cap costs on automated workflows without sacrificing quality on the specific steps that matter most.

Pricing Unchanged: $5/$25 Per Million Tokens

Despite the capability increase, Opus 4.7 maintains Opus 4.6 pricing: $5 per million input tokens and $25 per million output tokens. For teams already running agentic workflows on Opus 4.6, this is a straightforward upgrade with no budget impact.

Opus 4.7 is available across all Claude plans, the API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.

Claude Design: A New Anthropic Labs Product

Alongside the conference, Anthropic launched Claude Design — a new Labs product that lets you collaborate with Claude to create visual outputs: designs, prototypes, slides, and one-pagers. It shipped alongside Opus 4.7 and is positioned as a natural complement to Claude Code: Code handles the engineering, Design handles the visual layer.

This is early-stage, but the direction is clear. Anthropic is building a suite of specialized Claude-powered tools beyond the core chat interface, each optimized for a specific professional workflow.

Claude Security: Now in Public Beta

The other major announcement coinciding with the conference: Claude Security is now in public beta for Claude Enterprise customers. The feature allows security teams to scan code for vulnerabilities and generate proposed fixes using Opus 4.7 — either directly on the Claude platform or through partner integrations.

This positions Claude not just as a development accelerator but as an active participant in secure SDLC practices. For enterprise teams evaluating AI in regulated environments, the ability to run security scanning through the same model handling development work is a meaningful consolidation.

Why This Conference Signals a Shift

Step back and the pattern is clear. In 2025, "using AI for coding" meant tab completions and one-off questions. In 2026, the conversation has moved entirely: the question is no longer whether to use Claude in your development workflow, but how deep to integrate it.

Code with Claude exists because Anthropic's developer community has outgrown documentation. Engineers are deploying multi-agent systems, building custom MCP integrations, and running automated agentic loops in production — and they need hands-on guidance from the people who built the tools.

The three-city structure (SF → London → Tokyo) also signals something: this is a global developer story, not a Silicon Valley niche. The fact that the London date sold out and required a second day for indie developers and founders is particularly telling. It's not just enterprise teams — it's individual builders.

How to Attend or Watch

In-person registrations for San Francisco are closed (sold out). London (May 19–20) and Tokyo (June 10) registration is open at claude.com/code-with-claude. The full livestream for all city events is free and globally accessible. Session recordings are published after each event — so even if you miss the live broadcast, everything will be available.

If you're serious about building with Claude, watching the SF sessions is a direct shortcut to skills that would otherwise take weeks of trial and error to develop.

Key Takeaways

• Anthropic's Code with Claude conference kicked off May 6 in San Francisco, with London and Tokyo events to follow. It's structured around practical, hands-on technical workshops — not marketing talks.
• The core curriculum covers agentic coding with Claude Code, MCP implementation, and CLAUDE.md project configuration — the exact skills in demand for Claude Certified Architect exam candidates.
• Claude Opus 4.7 (launched April 16) is the technical centerpiece: high-resolution vision, effort levels, task budgets, and unchanged pricing make it a direct upgrade for agentic use cases.
• Claude Design and Claude Security both launched alongside the conference, expanding Anthropic's product suite beyond core chat.
• The conference is free to livestream globally — London and Tokyo in-person seats are still available.

Start Building with Claude Today

The skills being taught at Code with Claude — agentic workflows, MCP, Claude Code configuration — are exactly what the Claude Certified Architect exam tests. If you want a structured way to learn this material, our CCA practice test bank covers all exam domains with 200+ real-format questions.

Explore the CCA Practice Test Bank →

Or if you're brand new to Claude Code, our beginner's guide to Claude Code walks through the core setup in under 30 minutes.

---

Sources: Anthropic Code with Claude | Claude Opus 4.7 Release Notes | EventBrowse SF Event | TechFastForward Coverage