Skip to content

Self-evolving Context Database for AI Agents. Unify Agent Memory, Knowledge RAG and Skills.

1.6 KB
Raw Blame
metadata:
  id: "processing.interaction_learning"
  name: "Interaction Record Learning"
  description: "Extract reusable experience from interaction records"
  version: "1.0.0"
  language: "en"
  category: "processing"

variables:
  - name: "interactions_summary"
    type: "string"
    description: "Interaction summary"
    required: true

  - name: "effective_resources"
    type: "string"
    description: "Effective resource usage (contribution >= 0.3)"
    required: true

  - name: "successful_skills"
    type: "string"
    description: "Successful skill invocations"
    required: true

template: |
  Please analyze the following interaction records and extract reusable experience:

  Interaction Summary:
  {{ interactions_summary }}

  Effective Resource Usage (contribution >= 0.3):
  {{ effective_resources }}

  Successful Skill Invocations:
  {{ successful_skills }}

  Please extract the following types of experience:
  1. Resource usage insights: Which resources work well in which scenarios
  2. Skill combination patterns: Which skill combinations work well together
  3. User preferences: User's implicit preferences

  Output in JSON format:
  {
      "resource_insights": [
          {"resource_uri": "URI", "scenario": "Applicable scenario", "effectiveness": "Effectiveness description"}
      ],
      "skill_patterns": [
          {"skills": ["URI1", "URI2"], "scenario": "Applicable scenario", "strategy": "Usage strategy"}
      ],
      "user_preferences": [
          {"preference": "Preference description", "evidence": "Evidence"}
      ]
  }

llm_config:
  temperature: 0.0