Anthropic · 2026-11-06 · notable
Claude Code 2.1.319 — MCP tasks run in the background, deny instead of prompt
In short
Claude Code 2.1.319 moves MCP tool calls that run as tasks to the background, adds a setting that denies a tool call instead of waiting on a permission prompt, and tightens auto mode behind LLM gateways.
Anthropic · 2026-11-06 · 1 source checked
Claude Code 2.1.319 lets long MCP jobs finish in the background and lets unattended runs skip permission prompts.
What is it?
MCP tasks now work in the terminal in Claude Code 2.1.319: when an MCP server runs a tool call as a task, Claude Code moves it to the background right away and tells Claude the result when it finishes. A new `permissions.denyInsteadOfPrompting` setting denies any tool call that would wait on a permission prompt and lets the turn continue, in terminal sessions, subagents and teammates.
How does it work?
Auto mode changes behind LLM gateways: with telemetry or feature flags off, calls the server cannot check are asked about or denied instead of falling back to the built-in classifier, and `CLAUDE_CODE_AUTO_MODE_SERVER` switches that behavior. A new `CLAUDE_CODE_DISABLE_FEATURE_FLAGS` variable stops feature-flag fetching without turning off telemetry, and `/claude-api latency-optimize` looks for slow spots in your Claude API code and tests faster settings against your eval.
Why does it matter?
Scripts and background agents no longer stall on a prompt nobody will answer, and slow MCP jobs stop blocking the conversation. The release also closes several permission gaps, including a Bash bypass through a crafted environment prefix, a `sed` command that could skip the check, and a sandbox escape when creating a worktree.
Who is it for?
developers running Claude Code unattended, with MCP servers or behind a gateway
Try it
npm i -g @anthropic-ai/claude-code@2.1.319