TL;DR: Anthropic's Claude Code CLI update lets organizations centrally provision MCP servers and run unattended sessions without stuck permission prompts.
Summary: Claude Code 2.1.259 ships 37 CLI changes, notably managedMcpServers for org-wide provisioning of HTTP/SSE MCP servers to centralize tool distribution. It also adds --permission-prompts none for headless hosts to auto-deny prompts, and fixes concurrent sessions overwriting ~/.claude.json so workspace trust and MCP/project state persist.
Why it matters: AI builders can now standardize tool access for entire teams and safely run Claude Code in CI or background agents. Try rolling out org-managed MCP configs or enabling --permission-prompts none in automated pipelines.
Source: x_com