SDK Enhancements / Claude Code v2.1.119

Several new options are available for the Agent SDK:

askUserQuestion.previewFormat exposed via CLAUDE_CODE_QUESTION_PREVIEW_FORMAT env var
canUseTool callback for dynamic tool permission decisions (mutually exclusive with permissionPromptToolName)
getOAuthToken callback for custom OAuth token provision
Input/output validation with structured error messages: "Input validation error: Invalid arguments for tool" / "Output validation error: Invalid structured content for tool"
Tool annotations support via Zod schemas
taskSupport for tools: "required" (needs task augmentation) vs "forbidden" (must use registerTool() instead)

Evidence

SDK enhancements (search for "CLAUDE_CODE_QUESTION_PREVIEW_FORMAT", "canUseTool callback", "Output validation error")

Strings lifted out of the shipped bundle, so the claim above can be checked against them.

The build it came inv2.1.119

23 Apr 2026 · 35 entries, this one is #28.

Either side of it

Open the release →

Names in this entry

envCLAUDE_CODE_QUESTION_PREVIEW_FORMAT
Sets the AskUserQuestion preview format when the value is markdown or html. Otherwise markdown is chosen for clients other than SDK, claude-desktop, local-agent, remote and some VS Code sessions. The SDK sets this variable for the CLI process from its askUserQuestion.previewFormat option.
settingcanUseTool
Background and forked agents now run your permission callback even when a hook says allow, and deny rules win.
settingpermissionPromptToolName
The Agent SDK query builder accepts a permissionPrompts option of host or none, passed through to the CLI.
settingrequired
Plugins your administrator requires or auto-installs are now labelled as admin installs.

Is this right?