AI agent
Hooks status
The Kition runtime can run shell hooks on Agent events, but the app does not offer a Hooks settings pane or a hooks file to configure them.
Updated
By Kition Docs
Verified against Kition 0.2.0 (7d49036f) on 2026-11-06
Current product status
The local Kition runtime stores hook records and runs them on Agent events. Kition Settings does not include a Hooks pane, and no screen in the app lists, creates, or edits hooks.
There is no .kition/hooks.json or global hooks file. Old instructions that described a hooks file, project/global merge order, tool-argument mutation, or a blocking PreToolUse schema do not apply.
How the runtime treats hooks
- Agent events that can fire a hook: task started, task completed, task failed, tool called, and awaiting user input
- A hook record holds a name, event, optional tool/stage match pattern, command, working directory, timeout, and enabled state
- Commands run through the user’s shell; the working directory defaults to the workspace
- Timeouts must be between 1 and 300 seconds
- Hooks run in the background after the event; they cannot block, change, or approve a tool call
What not to do
- Do not create a hooks file from an old example and assume Kition will load it
- Do not assume a hook can mutate tool arguments or results
- Do not treat a hook as a replacement for shell approval, backups, or least-privilege credentials
- Do not place tokens directly in a hook command
If you rely on hooks
Treat each hook as local code execution with your account’s permissions. Pin the command, use a narrow working directory, keep timeouts short, and review what it does with its output before enabling it.
Implementation sources
Related pages
Agent approvals and safety boundaries
Use the shell-command approval card and kit drafts without assuming Kition has a universal per-tool permission editor.
Agent tool calls
Read the runtime tool names, how availability is decided, and the inputs, outputs, errors, and changed files shown in the Agent timeline.