Package com.tnsai.learning

Class AdaptivePromptEnricher

java.lang.Object

com.tnsai.learning.AdaptivePromptEnricher

public final class AdaptivePromptEnricher
extends Object

Enriches system prompts with learned rules from FeedbackLearner.

Closes the feedback-to-behavior loop: feedback recorded by users is analyzed by FeedbackLearner to extract actionable rules, which this class appends to the agent's system prompt before each LLM call.

Thread-safe. The enricher caches learning results and only re-learns when new feedback has been recorded since the last learning cycle.

Usage

 AdaptivePromptEnricher enricher = new AdaptivePromptEnricher(feedbackLearner, feedbackStore);
 String enrichedPrompt = enricher.enrich(basePrompt);
 

Since:

0.1.0

Constructor Summary

Constructors

Constructor	Description
AdaptivePromptEnricher(FeedbackLearner feedbackLearner, FeedbackStore feedbackStore)	Creates an enricher with the given feedback learner and store.

Method Summary

Modifier and Type	Method	Description
String	enrich(String basePrompt)	Enriches the base system prompt with learned rules and preferences.
List<String>	getLearnedPreferences()	Returns the current learned preferences.
List<String>	getLearnedRules()	Returns the current learned rules (prompt adjustments).
void	refresh()	Forces a re-learning cycle, bypassing the feedback count cache.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

AdaptivePromptEnricher

public AdaptivePromptEnricher(FeedbackLearner feedbackLearner,
 FeedbackStore feedbackStore)

Creates an enricher with the given feedback learner and store.

Parameters:

feedbackLearner - the learner that extracts rules from feedback

feedbackStore - the store that holds recorded feedback

Method Details

enrich

public String enrich(String basePrompt)

Enriches the base system prompt with learned rules and preferences.

If no rules have been learned (insufficient feedback or learner returned empty), returns the base prompt unchanged.

Parameters:

basePrompt - the original system prompt

Returns:

enriched prompt with learned behavior section appended, or the original prompt if no learnings are available

getLearnedRules

public List<String> getLearnedRules()

Returns the current learned rules (prompt adjustments).

Returns:

list of learned rules, never null

getLearnedPreferences

public List<String> getLearnedPreferences()

Returns the current learned preferences.

Returns:

list of learned preferences, never null

refresh

public void refresh()

Forces a re-learning cycle, bypassing the feedback count cache.