A daily digest of everything new about Claude and Anthropic

DeepSeek Is Building Its Own Claude Code. Beijing Wants the Whole Stack Decrypt [truncated: Google News RSS provides only a snippet, not full article

Anthropic's token pricing has escalated to levels causing enterprise concern, while open source LLM advances like Qwen 3.6 27B have made it increasingly feasible to run capable models locally on standard hardware with 16GB VRAM. Most coding tasks are characterized as intermediate difficulty rather than complex. The discussion suggests a potential shift toward free, locally-hosted coding agents as alternatives to paid cloud-based LLM services.

OpenAI's business model currently operates at a loss, with uncertainty about whether the market will tolerate such deep losses before an anticipated IPO. Anthropic is taking a different approach by establishing dominance in the enterprise market while pursuing a more conservative capacity strategy. These contrasting approaches represent two distinct pathways being tested in the artificial intelligence business landscape.

Dialogues with scholars, philosophers, clergy, and ethicists have been conducted to address questions raised by artificial intelligence. The conversations initiated with an exploration of how good character forms.

CLI and SDK maintainers can recommend plugins from the official Anthropic marketplace by writing a hint marker to stderr when the CLAUDECODE environment variable is detected. Claude Code reads the hint, removes it from output before sending to the model, and displays a one-time installation prompt to the user. The hint tag requires three attributes (version, type, and plugin identifier) and should be emitted at key moments such as help output or authentication success.

Anthropic has been organizing dialogues with scholars, clergy, and philosophers from religious and cultural traditions to inform the development of frontier AI systems, particularly regarding how AI character and values should be morally formed. The company experimented with giving Claude an ethical reminder tool that could be called mid-task, which resulted in markedly lower rates of misaligned behavior in internal evaluations. Plans for future engagement include conversations with legal scholars, psychologists, and civic institutions about broader questions surrounding AI's impact on work, institutions, and power distribution.

Claude Code and Codex projects can be unified by understanding their different file structures: Claude Code uses Claude.md and .claude folders while Codex uses agents.md, .codex, and .agents folders. The conversion process involves organizing shared knowledge accessible to any agent, skills and workflows stored in tool-specific locations, and tool-specific configurations, which can be automated by prompting an AI to create the necessary file structures. Maintaining dual compatibility requires keeping corresponding files synchronized when making updates to either version.

Claude Cowork's Small Business plugin enables multi-step automated workflows that integrate data from accounting, payment processing, CRM, and calendar tools to help manage business operations. Four example workflows—Monday business brief, campaign planning, month-end closing, and payroll planning—draft and stage work across existing business tools while keeping approval with the business owner before any action is taken. The plugin follows user permissions, customizations, and preferences to adapt recommendations to individual business needs.

Claude Skills are replacing prompt libraries as the standard development pattern due to superior portability, versioning, and automatic context-aware loading. Anthropic announced a major compute deal with SpaceX providing access to 220,000 Nvidia GPUs through Colossus 1, enabling doubled rate limits for Claude Code and increased API capacity. The author conducted 75 tests on Claude Skills to evaluate what breaks and provide an audit checklist for improving existing skills in production environments.

Aurora is a read-only connector that provides Claude access to Consilio engagement, document, and matter data, enabling full-text search and AI-powered investigations across multiple workspaces. The system automatically narrows queries to relevant context and cites source URLs for all records while maintaining access restrictions based on user entitlements.

Claude Opus 4.7 can audit large folders of visual assets against brand guidelines to identify violations including off-brand colors, outdated logos, and missing legal copy, providing categorized lists of violations with confidence ratings. The model reads images at higher resolution than previous versions, enabling detection of small details like specific hex codes and legal text, and processes hundreds of assets at once through Claude Cowork. Violations can be grouped by type and priority, integrated into project management systems like Asana, or scheduled for recurring audits with results posted to Slack.

Claude Code is an agentic coding environment that enables autonomous problem-solving by reading files, running commands, and implementing solutions rather than simply responding to questions. The primary constraint in using Claude Code effectively is managing the context window, which fills rapidly during coding sessions and causes performance degradation as it becomes full. Best practices for maximizing Claude Code's effectiveness include providing clear verification criteria, exploring and planning before implementation, offering specific contextual information in prompts, and establishing persistent instructions through a CLAUDE.md configuration file.

claude-cli — How Boris Uses Claude Code Boris created Claude Code at Anthropic. When asked how he uses it, he shared 13 practical tips from his daily workflow. His setup is "surprisingly vanilla" — proof that CC works great out of the box. Click the tabs

# Anthropic Courses Overview Anthropic offers a comprehensive suite of structured learning paths covering AI fundamentals and Claude development, including specialized tracks for agents (subagents, Claude Cowork, agent skills), the Claude API, and Model Context Protocol (MCP) at both introductory and advanced levels. Additionally, learners can find courses on AI fluency for different audiences (general and nonprofit-focused) plus deployment guides for Claude with major cloud providers (AWS Bedrock, Google Vertex AI). These courses combine video lessons and assessments to help developers build practical AI collaboration and integration skills.

Anthropic Academy provides a comprehensive development guide covering Claude API tools, resources, and educational materials for building with various Claude models. The guide encompasses tutorials, courses, and documentation for implementing the Messages API, File APIs, tool use, vision capabilities, RAG systems, prompt caching, MCP integration, and Claude Code. Code samples and implementation guides are available through the Anthropic Cookbook and Quickstarts repository for common development workflows.