AI & Behavior Traits

Part of the HoloScript Traits reference. Browse: Interaction · AI Autonomous · All Traits

AI/Behavior Traits

@llm_agent

Category: AI Tags: llm, agent, processing, autonomy, brain

LLM-powered decision-making with tool calling.

object NPC @llm_agent(model: 'gpt-4', temperature: 0.7) {
  system_prompt: "You are a helpful shopkeeper..."
  tools: ['get_inventory', 'sell_item', 'buy_item']
}

Config	Type	Default	Description
model	string	'gpt-4'	LLM model identifier
system_prompt	string	''	System message
temperature	number	0.7	Creativity (0-1)
context_window	number	8192	Max tokens
tools	Array	[]	Available tool definitions
max_actions_per_turn	number	3	Action limit per turn
bounded_autonomy	boolean	true	Limit autonomous actions
escalation_conditions	Array	[]	When to escalate
rate_limit_ms	number	1000	Min time between requests

State:

• conversationHistory - Message history
• isProcessing - Currently processing
• pendingToolCalls - Pending tool invocations

Events:

• llm_message - Message received
• llm_response - Response generated
• llm_tool_call - Tool invoked
• llm_escalate - Escalation triggered

---

@behavior_tree

Category: AI Tags: bt, logic, decision, npc, state

Behavior tree AI with node-based logic.

object Guard @behavior_tree {
  root: sequence [
    condition 'can_see_player',
    action 'chase_player',
    action 'attack'
  ]
}

Config	Type	Default	Description
tick_rate	number	10	Updates per second
debug_mode	boolean	false	Show node execution
interrupt_on_higher_priority	boolean	true	Priority interrupts

Events:

• bt_node_enter - Node started
• bt_node_success - Node succeeded
• bt_node_failure - Node failed
• bt_tree_complete - Full tree evaluated

---

@goal_oriented

Goal-oriented action planning (GOAP).

object Worker @goal_oriented {
  goals: ['gather_resources', 'build_shelter']
  actions: ['chop_wood', 'mine_stone', 'craft']
}

Config	Type	Default	Description
planning_depth	number	5	Max action chain length
replan_interval	number	1000	Replanning frequency (ms)

Events:

• goal_selected - New goal chosen
• goal_achieved - Goal completed
• action_started - Action begun
• action_completed - Action finished

---

@perception

Category: AI Tags: vision, hearing, sense, detection, npc

Sensory perception with line-of-sight and detection.

object Sentry @perception(view_angle: 90, view_distance: 20) {
  detect_tags: ['player', 'enemy']
}

Config	Type	Default	Description
view_angle	number	120	Field of view (degrees)
view_distance	number	10	Max sight distance
hearing_radius	number	5	Sound detection radius
memory_duration	number	5000	How long to remember (ms)

State:

• visibleTargets - Currently visible entities
• heardSounds - Recent sounds detected
• memory - Remembered but not visible

Events:

• perception_spotted - Target spotted
• perception_lost - Target lost
• perception_heard - Sound detected

---

@emotion

Emotional state machine with expressions.

object Pet @emotion {
  emotions: ['happy', 'sad', 'angry', 'neutral']
  default_emotion: 'neutral'
}

Config	Type	Default	Description
decay_rate	number	0.1	How fast emotions fade
expression_blend_time	number	500	Blend duration (ms)

State:

• currentEmotion - Active emotion
• emotionIntensity - Intensity (0-1)
• emotionHistory - Recent emotions

Events:

• emotion_changed - Emotion transitioned
• emotion_triggered - Emotion stimulus received

---

@memory

Long-term memory and recall for NPCs.

object Historian @memory(capacity: 100, consolidation_interval: 60000) {
  memory_types: ['events', 'conversations', 'locations']
}

Config	Type	Default	Description
capacity	number	100	Max memories stored
consolidation_interval	number	60000	Memory cleanup (ms)
importance_threshold	number	0.3	Min importance to keep

Events:

• memory_store - Memory saved
• memory_recall - Memory retrieved
• memory_forget - Memory discarded

---

See Also

• AI Autonomous Traits
• MCP Server Integration
• API Reference