Skip to main content

Interface: SentimentTrackingConfig

Defined in: packages/agentos/src/cognition/substrate/personas/IPersonaDefinition.ts:259

Configuration for sentiment-aware metaprompt tracking. Controls whether the GMI analyzes user sentiment and triggers event-based metaprompts in response to detected emotional states.

Interface​

SentimentTrackingConfig

Examples​

// Minimal: enable with defaults
sentimentTracking: { enabled: true }
// Full: custom thresholds; 'llm' sends a HybridUtilityAI's scoring to its LLM-backed class
sentimentTracking: {
  enabled: true,
  method: 'llm',
  historyWindow: 10,
  frustrationThreshold: -0.3,
  satisfactionThreshold: 0.3,
  consecutiveTurnsForTrigger: 2,
  presets: ['frustration_recovery', 'confusion_clarification'],
}

Properties​

consecutiveTurnsForTrigger?​

optional consecutiveTurnsForTrigger: number

Defined in: packages/agentos/src/cognition/substrate/personas/IPersonaDefinition.ts:314

Number of consecutive turns with same sentiment pattern before triggering event. Prevents over-triggering on single outlier messages.

Default​

2

enabled​

enabled: boolean

Defined in: packages/agentos/src/cognition/substrate/personas/IPersonaDefinition.ts:266

Master switch: enables/disables sentiment analysis on user input. When false (default), no sentiment analysis runs and no events are emitted. Turn_interval metaprompts (like gmi_self_trait_adjustment) still work regardless.

Default​

false

frustrationThreshold?​

optional frustrationThreshold: number

Defined in: packages/agentos/src/cognition/substrate/personas/IPersonaDefinition.ts:301

Sentiment score threshold below which frustration is detected. Score range: -1 (very negative) to 1 (very positive).

Default​

-0.3

historyWindow?​

optional historyWindow: number

Defined in: packages/agentos/src/cognition/substrate/personas/IPersonaDefinition.ts:294

Number of recent turns to keep in sentiment history (sliding window). Higher = better pattern detection, slightly more memory.

Default​

10

method?​

optional method: "llm" | "lexicon_based" | "trained_classifier"

Defined in: packages/agentos/src/cognition/substrate/personas/IPersonaDefinition.ts:275

The method the tracker passes to the GMI's utility class with each turn. The utility class decides how a turn is scored: LLMUtilityAI makes a provider call for each turn (and a repair call when its reply does not parse) and StatisticalUtilityAI scores with a word lexicon and makes none, whatever this names. A HybridUtilityAI that holds both sends 'llm' to its LLM-backed class and 'lexicon_based' or 'trained_classifier' to its statistical class.

Default​

'lexicon_based'

modelId?​

optional modelId: string

Defined in: packages/agentos/src/cognition/substrate/personas/IPersonaDefinition.ts:281

The modelId an LLMUtilityAI scores a turn with, else the persona's defaultModelId. StatisticalUtilityAI does not read it.

presets?​

optional presets: ("all" | "frustration_recovery" | "confusion_clarification" | "satisfaction_reinforcement" | "error_recovery" | "engagement_boost")[]

Defined in: packages/agentos/src/cognition/substrate/personas/IPersonaDefinition.ts:328

Which preset metaprompts to enable. Options:

  • 'frustration_recovery': Responds to user frustration
  • 'confusion_clarification': Responds to user confusion
  • 'satisfaction_reinforcement': Responds to user satisfaction
  • 'error_recovery': Responds to error accumulation
  • 'engagement_boost': Responds to low engagement
  • 'all': Enables all presets

Only listed presets will be merged. Omit to enable none (use custom metaPrompts instead).

Default​

[] (no presets auto-merged)

providerId?​

optional providerId: string

Defined in: packages/agentos/src/cognition/substrate/personas/IPersonaDefinition.ts:287

Provider ID for LLM-based methods. Falls back to persona defaultProviderId if not specified.

satisfactionThreshold?​

optional satisfactionThreshold: number

Defined in: packages/agentos/src/cognition/substrate/personas/IPersonaDefinition.ts:307

Sentiment score threshold above which satisfaction is detected.

Default​

0.3