flowai-skill-engineer-hook

To configure agent hooks in Claude Code, you wire a new hook entry into a specific event using the provided IDE references. This allows you to intercept, evaluate, and apply rules to block dangerous commands or enforce coding standards during automated agent interactions.

Yes, you can use a unified approach for creating and configuring hooks across Cursor, Claude Code, and OpenCode. The skill provides IDE detection, 4 hook types, 18 events mapping, and example templates to ensure consistent governance and safety checks across these different environments.

Agent hooks are event plugins that manage agent behavior, commands, file reads/writes, and tool usage within IDEs. You need them to apply consistent governance, auditing, and safety checks, allowing you to allow, block, request confirmation, or modify input data during automated workflows.

You create a new hook entry by wiring a new hook into an event using the IDE references and the provided shell script template. This process enables you to script custom logic to intercept, evaluate, and modify input data for file writes or other agent actions in Cursor.

Yes, the hook configuration supports scripting logic to modify input data before an action executes. Alongside allowing or blocking actions with an explanation, you can intercept and evaluate agent commands or file operations to dynamically alter their input data across supported IDE environments.