Skip to main content

hooks.base

Hook event types and data models.

HookEvent Objects​

class HookEvent(str, Enum)

Events that can trigger hooks.

PRE_TOOL_USE​

Before tool execution. Can allow, deny, or modify arguments.

POST_TOOL_USE​

After successful execution. Can modify the result.

POST_TOOL_USE_FAILURE​

After failed execution. For error reporting/recovery.

HookInput Objects​

@dataclass
class HookInput()

Input passed to a hook handler.

Parameters​

event : HookEvent The event that triggered this hook. tool_name : str Name of the tool being executed. tool_args : dict Tool call arguments (for PRE hooks, may be modified). tool_result : str | None Tool execution result (POST hooks only). error : Exception | None Tool execution error (POST_TOOL_USE_FAILURE only). agent_id : str | None The agent ID. user_id : str | None The requesting user ID. metadata : dict Additional context (turn number, token count, etc.).

HookResult Objects​

@dataclass
class HookResult()

Result returned from a hook handler.

Parameters​

allow : bool PRE_TOOL_USE: False → deny the tool call. reason : str | None Explanation (shown to agent if denied). modified_args : dict | None PRE: Replace tool arguments. modified_result : str | None POST: Replace tool output.