🧭 Claude.ai Projects — Shared Team Workspaces Now Available

Claude.ai Teams and Enterprise users can now create shared Projects — persistent workspaces that multiple team members can access, contribute to, and build on together. Previously, Projects were private to the individual user who created them. The update makes shared Projects the foundation for teams that want to maintain a living set of instructions, reference documents, and conversation history around a recurring task or domain.

How shared Projects work

• Shared knowledge — documents uploaded to a shared Project are available to all members; Claude draws on them in every conversation started within the Project
• Shared CLAUDE.md — a single project-level instruction file governs Claude's behaviour for all team members, ensuring consistent output style and constraints
• Conversation visibility settings — conversations within a shared Project can be set to private, team-visible, or workspace-visible, giving members control over what they share
• Activity log — workspace admins can see Project membership and document changes in the audit log

Shared Projects are available today for Teams and Enterprise plan customers. Anthropic notes that this is the first step toward richer collaboration features planned for 2026, including real-time shared editing and conversation branching.

🧭 MCP Directory Passes 200 Active Community Servers

The Model Context Protocol public server directory has passed 200 actively maintained listings — a milestone that underscores the pace of community adoption since MCP was published as an open standard in late 2024. The 200-server mark was reached six months ahead of Anthropic's internal projections for the ecosystem at this stage.

The distribution across categories reflects the practical priorities of developers building AI-connected tools: developer toolchains (GitHub, GitLab, Linear, Jira integrations) represent the largest cluster, followed by document and knowledge management (Notion, Confluence, Google Drive), and data access (Postgres, BigQuery, Snowflake). A growing long-tail category covers scientific and domain-specific databases — PubMed, the arXiv API, financial data providers — that developers have been quietly wiring up for research-oriented Claude workflows.