Claude Code 2.1.248
By Steven Van ·
Added --restricted (or CLAUDE_CODE_RESTRICTED=1): removes the built-in tools that run commands or code and WebFetch (unless named in --tools), keeps file…
Claude Code 2.1.248 adds a locked-down mode for running the tool in less trusted environments, along with several smaller changes for enterprise admins and agent authors, detailed in the changelog.
- A new --restricted flag (or CLAUDE_CODE_RESTRICTED=1) removes the built-in tools that run commands or code and WebFetch, unless those tools are explicitly named in --tools. It also keeps file tools confined to the working directory, refuses bypassPermissions, and ignores user, project and local settings files.
- Agent frontmatter can now set experimental.cacheTtl to "5m" or "1h", giving each agent its own prompt cache TTL when no subagent TTL setting is configured.
- claude self-hosted-runner gains a --client-label flag (or SELF_HOSTED_RUNNER_CLIENT_LABEL) to override the label the runner registers with, which otherwise defaults to the hostname.
- Server-managed settings now get diagnostics: a startup warning when they fail to load, and a line in /doctor and /status explaining a load failure or why they weren't fetched, covering Bedrock, Vertex, third-party providers and custom ANTHROPIC_BASE_URL setups.
- /web-setup now warns when the GitHub CLI token lacks the workflow scope, since pushes to very large repositories can otherwise be rejected.
- /usage-credits lets members of Enterprise organizations billed through AWS Marketplace, self-serve Enterprise and Enterprise trials request a higher usage limit from their admin.
- Agents can now message across sessions via SendMessage and ListAgents.
Claude Code
Anthropic's agentic coding tool — Claude in your terminal, IDE, desktop app and browser, reading and editing your codebase, running commands, opening pull requests and working on tasks in the cloud.
View Claude Code →