Scaffold all 7 hook events and eliminate type casting

- Add missing hook handlers: Stop, SubagentStop, PreCompact
- Replace all string type casting with proper JSON unmarshaling
- Change ToolUseRequest.Parameters from map[string]interface{} to json.RawMessage
- Update helper methods to use json.Unmarshal directly to typed structs
- Replace speech notifications with simple logging notifications
- Update documentation to reflect all hook events and remove speech references
- Add comprehensive CLAUDE.md with architecture overview and development guidance

All hooks now use type-safe JSON handling without .(string) assertions.
Complete coverage of Claude Code events: PreToolUse, PostToolUse,
UserPromptSubmit, Notification, Stop, SubagentStop, PreCompact.

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

internal/handlers/subagentstop.go

@@ -0,0 +1,42 @@
+package handlers
+
+import (
+	"context"
+	"log/slog"
+	"time"
+
+	"github.com/imjasonh/hooks/internal/hooks"
+)
+
+func init() {
+	hooks.RegisterHook(hooks.EventSubagentStop, "*", LogSubagentCompletion)
+}
+
+// LogSubagentCompletion is triggered when a Claude Code subagent (Task tool call)
+// finishes responding. Subagents are created when Claude uses the Task tool to
+// delegate work to another Claude instance.
+//
+// Common use cases:
+// - Track nested workflow completion
+// - Monitor subagent performance
+// - Resource cleanup after Task operations
+// - Debug complex multi-agent workflows
+// - Collect metrics on task delegation
+func LogSubagentCompletion(ctx context.Context, input hooks.HookInput) (hooks.HookOutput, error) {
+	slog.Info("Claude subagent completed",
+		"session_id", input.SessionID,
+		"cwd", input.CWD,
+		"transcript_path", input.TranscriptPath,
+		"timestamp", time.Now().Unix())
+
+	// TODO: Future enhancements could include:
+	// - Track subagent hierarchy depth
+	// - Measure subagent execution time
+	// - Collect subagent success/failure rates
+	// - Clean up subagent-specific resources
+	// - Aggregate results from multiple subagents
+	// - Send notifications for long-running subagents
+	// - Export subagent conversation logs
+
+	return hooks.HookOutput{Decision: "approve"}, nil
+}