hook-development

PreToolUse hooks in Claude Code can be configured as either prompt-based or command hooks to validate tool inputs. You define them in your plugin setup to intercept and check parameters before execution.

Prompt-based hooks use natural language instructions for validation, while command hooks execute shell scripts with dependencies like jq or curl. Both patterns support event-driven automation for PreToolUse and PostToolUse workflows.

Debugging Claude Code plugin hooks involves checking your JSON output formats, verifying timeout configurations, and tracing event triggers. The framework provides specific debugging workflows to identify and resolve hook execution failures.

Yes, command hooks often require dependencies like jq, curl, psql, md5sum, and nc to execute shell-based validation and automation. Ensure these tools are installed in your environment for reliable hook performance.

PostToolUse hooks enable safety checks and validation after tool execution completes. You can implement command or prompt-based patterns to verify outputs and ensure secure automation around tool usage events.

Claude Code hook security guidelines specify configuring strict timeouts, validating JSON outputs, and limiting command hook permissions. Following these ensures reliable, secure, and maintainable plugin hook implementations.