Claude Code 2.1.283
By Steven Van ·
Claude Code 2.1.283 adds gateway hint headers, model allow/deny controls for admins, and a CLAUDE.md prompt audit command.
- Requests can now carry an x-claude-code-prompt-id header so LLM gateways can group all the requests that come from one user prompt; it's opt-in via CLAUDE_CODE_GATEWAY_HINT_HEADERS=1.
- Admins get two new managed settings: availableModelsMatch set to "exact" locks an availableModels entry to that specific model version, so new releases stay blocked until explicitly listed, and deniedModels blocks named models even if availableModels would otherwise allow them.
- With OTEL_LOG_TOOL_CONTENT=1, the tool.output OpenTelemetry span event now also captures MCP tool, WebFetch and WebSearch outputs.
- A new /doctor prompt-audit command (also available as /checkup prompt-audit) audits CLAUDE.md files, skills, agents and commands for prompting patterns written for older models.
- In fullscreen mode, truncated messages from other sessions can now be clicked to expand.
- When a plugin directory fails to load, the --plugin-dir load-failure entry in the stream-json system/init output now names the directory path.
- The Claude apps gateway config adds an opt-in load_test_mode block: requests are built and signed but not sent upstream, and clients get a canned reply, allowing a deployment to be load tested.
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 →