Claude Report — 2026-06-21
- Enterprise-Managed MCP Connector Access launches in beta (June 18): Okta-provisioned zero-touch access to Asana, Figma, Linear, and 4 other connectors for Team/Enterprise orgs
- Claude Code Artifacts (June 18): sessions now output live, version-tracked shareable web pages, in beta for Team and Enterprise
- Claude Code v2.1.183 (June 19) adds auto-mode guardrails blocking destructive git operations and IaC teardown commands unless explicitly requested
- GitHub MCP Server v1.4.0 (June 18) defaults all new repository creation to private
- Claude Code v2.1.185 (June 20) improves connection-stall messaging and raises silence threshold to 20s before retry
🚀 Anthropic Official
Enterprise-Managed MCP Connector Access — Okta beta (June 18)
Anthropic launched Enterprise-Managed Authorization (EMA) for MCP connectors in beta for Team and Enterprise plans: admins provision connectors once via their Okta identity provider and every employee inherits access automatically on first login, with no individual OAuth consent screens. Connectors available at launch: Asana, Atlassian, Canva, Figma, Granola, Linear, and Supabase; authorization covers Claude chat, Claude Code, and Cowork in a single policy.
Claude Code Artifacts (June 18)
Claude Code sessions can now generate live, shareable web pages ("Artifacts") built from the full session context — codebase, connected services, and conversation. Artifacts update in place, carry version history, and are private by default with org-only sharing; available in beta via CLI and desktop for Team and Enterprise organizations.
Claude Code v2.1.183 (June 19)
Auto mode now blocks destructive git commands (reset --hard, checkout -- ., clean -fd, stash drop) unless the user explicitly asked to discard local work; git commit --amend is blocked for commits not created by the agent this session; and terraform destroy, pulumi destroy, cdk destroy are blocked unless the specific stack was named in the request. Adds model deprecation warnings on stderr and /config --help to list all shorthand configuration keys. Fixes: thinking-only API responses that silently completed a turn (Claude now re-prompts), WebSearch returning empty results inside subagents, fullscreen TUI corruption under heavy nested-subagent load in Windows Terminal, and subagent spawns now pass through the auto-mode safety classifier before launch.
Claude Code v2.1.185 (June 20)
Renames the stream-stall indicator from "No response from API · Retrying in …" to "Waiting for API response · will retry in …" and extends the silence window before auto-retry fires from 10 seconds to 20 seconds, reducing false-positive reconnection attempts on slow but healthy connections.
🔌 Claude Code Plugins
Plugin ranking cache unavailable this cycle — no feature-level plugin updates confirmed in the window.
Unchanged in window: tracking skipped — plugin ranking cache unavailable this cycle
🛠️ Skills
None of the top 5 updated in window.
Unchanged in window: Piebald-AI/claude-code-system-prompts, anthropics/knowledge-work-plugins, glebis/claude-skills, anthropics/skills, ComposioHQ/awesome-claude-skills
🤖 Agents & Subagents
claude-agent-sdk-python v0.2.105 + v0.2.106 (June 19 & 20)
Two back-to-back releases bundle Claude CLI v2.1.183 (June 19 drop) and v2.1.185 (June 20 drop) respectively; all auto-mode guardrails, deprecation warnings, /config --help, and connection-stall improvements from those CLI versions land automatically on pip install --upgrade claude-agent-sdk.
Unchanged in window: claude-agent-sdk-typescript (no update confirmed), nicobailon/pi-subagents, Chachamaru127/claude-code-harness
🔗 MCPs & Integrations
github/github-mcp-server v1.4.0 (June 18)
Defaults repository creation to private when visibility is omitted; adds an explicit show_ui parameter to UI-enabled write tools; introduces a get_commits method scoped to pull requests; ships a new tool for retrieving code quality findings; and supports custom listen addresses in HTTP mode.
Unchanged in window: modelcontextprotocol/python-sdk, modelcontextprotocol/typescript-sdk, awslabs/mcp, portainer/portainer-mcp
💡 Community — Workflows & Ideas
No relevant community posts with verified publication dates in the window.
📰 Quick Mentions
Relevant links that don't warrant a section of their own:
- TechTimes — "MCP Enterprise Authorization Goes Stable: Zero-Touch SSO for Okta, Anthropic, VS Code" (June 19) — broader industry framing for the Anthropic/Okta EMA launch, including VS Code context
- Okta Newsroom — "Okta becomes a featured identity provider powering secure AI agent connections for Anthropic's Claude" — Okta's perspective on the partnership and connector governance model
- TechTimes — "Claude MCP Connectors Now Provision Through Okta: Employees Inherit Access on Login" (June 19) — detailed walkthrough of the zero-touch provisioning flow from an IT administrator perspective
- Releasebot — Anthropic June 2026 — automated feed tracking all Anthropic releases this month