Reorganize repo: backend/, deploy/, docs/ layout + AGENTS.md

Restructure the monorepo into clear top-level directories:
- backend/: Go module root (cmd, internal, pkg, configs, migrations,
  docs/swagger, scripts, tests, go.mod, Makefile, .air.toml)
- deploy/: Docker (Dockerfile, docker-compose*), quickstart, fluentd
- docs/: project documentation + reports/ (moved from repo root)
- AGENTS.md: new AI coding-agent guide at repo root

Update all references to the new layout:
- Dockerfile: COPY backend/go.mod, COPY backend/ (context = repo root)
- docker-compose files: context ../.., dockerfile deploy/docker/Dockerfile,
  env_file ../../.env, volume mounts ../../backend:/app
- deploy/quickstart/compose.yaml: dockerfile deploy/docker/Dockerfile
- CI: working-directory: backend for go commands, file deploy/docker/Dockerfile,
  coverage path backend/coverage.out, health_check backend/scripts/
- backend/Makefile: docker target uses -f ../deploy/docker/Dockerfile ../
- README: architecture tree, quickstart, config paths updated

Move root stray scripts (rename_models.*, run_m11_tests.sh, verify_build.sh,
gorm_bool_main.go) to backend/scripts/legacy/. All moves via git mv to
preserve history. Build, vet, SQLite tests, and docker compose config verified.

backend/internal/model/auto_reply_rule_models.go

@@ -0,0 +1,79 @@
+package model
+
+import (
+	"encoding/json"
+)
+
+// --- Auto-Reply Rule Model ---
+// Reference: M12 PRD §Captain AI — Auto-Reply Rules
+// Table: captain_auto_reply_rules
+//
+// AutoReplyRule defines conditions and responses for automated replies.
+// When a conversation matches the rule conditions, Captain can auto-send
+// a predefined response or use LLM to generate a contextual reply.
+
+type AutoReplyRuleStatus string
+
+const (
+	AutoReplyRuleStatusActive  AutoReplyRuleStatus = "active"
+	AutoReplyRuleStatusDraft   AutoReplyRuleStatus = "draft"
+	AutoReplyRuleStatusPaused  AutoReplyRuleStatus = "paused"
+)
+
+type AutoReplyRuleMode string
+
+const (
+	AutoReplyRuleModeStatic AutoReplyRuleMode = "static"  // predefined response text
+	AutoReplyRuleModeLLM    AutoReplyRuleMode = "llm"     // LLM-generated contextual response
+	AutoReplyRuleModeMixed  AutoReplyRuleMode = "mixed"   // static intro + LLM contextual body
+)
+
+type CaptainAutoReplyRule struct {
+	Base
+	AccountID    uint               `gorm:"index;not null" json:"account_id"`
+	AssistantID  uint               `gorm:"index;not null" json:"assistant_id"`
+	InboxID      *uint              `gorm:"index" json:"inbox_id,omitempty"` // optional: scope to specific inbox
+	Name         string             `gorm:"size:255;not null" json:"name"`
+	Description  string             `gorm:"type:text" json:"description,omitempty"`
+	Status       AutoReplyRuleStatus `gorm:"size:50;default:draft;not null" json:"status"`
+	Mode         AutoReplyRuleMode  `gorm:"size:50;default:static;not null" json:"mode"`
+	Priority     int                `gorm:"default:0;not null" json:"priority"` // higher = evaluated first
+
+	// Conditions: JSONB array of condition objects that must all match
+	// Each condition: {"field": "message_content", "operator": "contains", "value": "urgent"}
+	// Fields: message_content, sender_type, conversation_status, language, keywords
+	Conditions json.RawMessage `gorm:"type:jsonb;serializer:json" json:"conditions"`
+
+	// Response config for static/mixed mode
+	ResponseText string `gorm:"type:text" json:"response_text,omitempty"` // static reply text
+
+	// LLM prompt override for llm/mixed mode
+	LLMPromptOverride string `gorm:"type:text" json:"llm_prompt_override,omitempty"`
+
+	// Delay before auto-reply (seconds, 0 = immediate)
+	DelaySeconds int `gorm:"default:0;not null" json:"delay_seconds"`
+
+	// Only auto-reply once per conversation
+	OneTimeOnly bool `gorm:"default:true;not null" json:"one_time_only"`
+}
+
+func (CaptainAutoReplyRule) TableName() string { return "captain_auto_reply_rules" }
+
+// AutoReplyCondition represents a single matching condition.
+type AutoReplyCondition struct {
+	Field    string `json:"field"`    // "message_content", "sender_type", "language", "keywords"
+	Operator string `json:"operator"` // "contains", "equals", "regex", "starts_with", "language_is"
+	Value    string `json:"value"`    // the value to match against
+}
+
+// GetConditions unmarshals the conditions JSON.
+func (r *CaptainAutoReplyRule) GetConditions() ([]AutoReplyCondition, error) {
+	if len(r.Conditions) == 0 || string(r.Conditions) == "null" || string(r.Conditions) == "[]" {
+		return nil, nil
+	}
+	var conditions []AutoReplyCondition
+	if err := json.Unmarshal(r.Conditions, &conditions); err != nil {
+		return nil, err
+	}
+	return conditions, nil
+}