A Friday that moved on two tracks. On the product side: Claude Code Artifacts ships in beta for Team and Enterprise accounts, turning any agent session into a live, interactive web page that updates as the agent works, shareable within the org via URL. On the infrastructure side: Enterprise-Managed Authorization for MCP connectors stabilizes, meaning IT admins provision access once through Okta and employees inherit every approved connector on first login with no OAuth queue.
Both features land for the same tier. Both are in beta. Two bets landing the same day.
git reset --hard, git checkout -- ., git clean -fd, git stash drop) unless you explicitly asked to discard local work. git commit --amend is blocked for commits the agent did not make this session. Terraform, Pulumi, and CDK destroy commands require explicit request per stack. Model deprecation warnings now appear in print mode (-p) and for models set in agent frontmatter.claude update or reinstall). Safety gates are automatic in auto mode. New setting: attribution.sessionUrl (set to false to omit the claude.ai session link from commits and PRs in web and Remote Control sessions). Run /config --help to list available shorthand config keys.fallback-refusal-middleware marker, making it easier to trace which requests hit the fallback path.pip install anthropic==0.111.0code_execution_20260120 tool. Fixes x-stainless-helper header merging (no longer clobbers across multiple header merge calls). Preserves Bedrock stream event types correctly.pip install anthropic==0.110.0 · Use code_execution_20260120 as the tool type in your tool definitions.code_execution_20260120 tool. Adds lazy parsing of partial tool JSON input, reducing overhead during streaming tool calls.npm install @anthropic-ai/sdk@0.105.0Daily digest at 9 PM ET. Weekly magazine every Friday morning. Six labs, one feed. No spam, one-click unsubscribe.