# Clustering Configuration for Insight-to-Skill Generator
# Version: 0.1.0

# Similarity Scoring Weights
similarity_weights:
  same_category: 0.3       # Base score if insights are in same category
  shared_keyword: 0.1      # Added per shared keyword
  temporal_proximity: 0.05 # If insights created within temporal_window
  title_similarity: 0.15   # If titles share significant words
  content_overlap: 0.2     # If content has overlapping concepts

# Clustering Thresholds
thresholds:
  cluster_minimum: 0.6     # Minimum score to group insights together
  standalone_quality: 0.8  # Score for single insight to become standalone skill
  split_cluster_size: 5    # If cluster > this size, consider splitting by sub-topics

# Temporal Settings
temporal_window: 7         # Days - insights within this window get proximity bonus

# Category Keywords (from extract-explanatory-insights hook)
# Used for semantic grouping beyond directory structure
categories:
  testing:
    keywords:
      - test
      - testing
      - coverage
      - tdd
      - unit
      - integration
      - e2e
      - vitest
      - jest
      - assertion
      - mock
      - fixture
    skill_category: tooling

  configuration:
    keywords:
      - config
      - configuration
      - settings
      - inheritance
      - yaml
      - json
      - env
      - environment
    skill_category: tooling

  hooks-and-events:
    keywords:
      - hook
      - lifecycle
      - event
      - trigger
      - callback
      - listener
      - middleware
    skill_category: productivity

  security:
    keywords:
      - security
      - vulnerability
      - auth
      - authentication
      - authorization
      - encryption
      - sanitize
      - injection
      - xss
      - csrf
    skill_category: analysis

  performance:
    keywords:
      - performance
      - optimize
      - optimization
      - cache
      - caching
      - lazy
      - memoize
      - bundle
      - latency
      - throughput
    skill_category: productivity

  architecture:
    keywords:
      - architecture
      - design
      - pattern
      - structure
      - module
      - component
      - layer
      - separation
      - coupling
      - cohesion
    skill_category: analysis

  version-control:
    keywords:
      - git
      - commit
      - branch
      - merge
      - rebase
      - pull request
      - pr
      - cherry-pick
    skill_category: devops

  react:
    keywords:
      - react
      - component
      - tsx
      - jsx
      - hooks
      - useEffect
      - useState
      - props
      - state
      - render
    skill_category: tooling

  typescript:
    keywords:
      - typescript
      - type
      - interface
      - generic
      - enum
      - type guard
      - union
      - intersection
    skill_category: tooling

  general:
    keywords:
      - general
      - best practice
      - lesson
      - debugging
      - troubleshooting
    skill_category: productivity

# Complexity Assessment Rules
# Determines whether insight(s) become minimal/standard/complex skill
complexity_rules:
  minimal:
    max_insights: 1
    max_paragraphs: 3
    has_code_examples: false
    description: "Single focused insight, phase-based workflow"

  standard:
    max_insights: 4
    min_paragraphs: 3
    requires_data_dir: true
    description: "Multiple related insights, comprehensive workflow with reference materials"

  complex:
    min_insights: 5
    requires_modes: true
    requires_examples: true
    description: "Large insight cluster, mode-based with extensive examples and templates"

# Skill Naming Heuristics
naming:
  max_length: 40           # Max chars for skill name (kebab-case)
  remove_words:            # Common words to remove from auto-generated names
    - "insight"
    - "lesson"
    - "learned"
    - "the"
    - "a"
    - "an"
  preferred_suffixes:      # Preferred endings for skill names
    - "guide"
    - "advisor"
    - "helper"
    - "automation"
    - "analyzer"

# Description Generation
description:
  max_length: 150          # Soft limit for description (actual limit is 1024)
  required_elements:
    - action_verb          # Must start with verb (e.g., "Use", "Analyzes", "Guides")
    - trigger_phrase       # Must include "PROACTIVELY when" or "Use when"
    - benefit              # Must describe value/outcome
  action_verbs:
    - "Use PROACTIVELY when"
    - "Guides"
    - "Analyzes"
    - "Automates"
    - "Validates"
    - "Optimizes"
    - "Generates"
    - "Monitors"