Claude Code 2.1.290
By Steven Van ·
Plugin authors get richer permission-check data, while terminal users can attach to sessions using part of a name.
Plugin hooks gain more context for tool calls and permission checks in Claude Code 2.1.290. The release also adds session-name matching and a way to deny pending sign-ins.
Hook data and validation
- A mod’s turn.step hook result now includes serverToolUses for tool calls the API ran itself through the advisor, with each call’s id, name, input, start and end.
- Plugin tool.check events now include agentId, allowing hooks to distinguish a subagent’s permission check from the main session’s.
- The question and verdict read by a mod’s tool.check hook now include ceiling, naming the approval an organization requires for a tool.
- Plugin hook typings add ThemeKey and Color, so editors can list the theme colors a mod’s drawing can name.
- claude plugin validate now lists each hook a mod registers at a gating site and whether it has a .catch. JSON output includes this under gatingHooks.
Sign-in and session commands
The Claude apps gateway’s sign-in approval page now has a Deny button. It ends the pending sign-in, stopping the waiting terminal within seconds. The claude attach and claude logs commands also accept part of a session name in place of its id.
Managed Agents setup
The /claude-api managed-agents-onboard command can set up a Managed Agents pattern from a URL or build a Console quickstart template, such as deep-researcher, with the ant CLI. Claude Code also warns when a managed settings file links to a file outside its folder.
