proagents 1.6.20 → 1.6.21

package/.proagents/learning/adaptation.md

@@ -1,305 +0,0 @@
-# Learning Adaptation
-
-How ProAgents applies learned patterns to improve over time.
-
----
-
-## Adaptation Types
-
-### 1. Automatic Corrections
-
-When the same correction is made 3+ times, apply automatically:
-
-```yaml
-auto_corrections:
-  threshold: 3  # Apply after 3 similar corrections
-
-  example:
-    pattern: "AI suggests Redux"
-    correction: "User chooses Zustand"
-    occurrences: 5
-    status: "auto_applied"
-    action: "Now suggests Zustand by default"
-```
-
-**Adaptation Flow:**
-```
-Correction 1: "Use Zustand, not Redux" → Noted
-Correction 2: "Use Zustand, not Redux" → Noted
-Correction 3: "Use Zustand, not Redux" → AUTO-APPLY ENABLED
-Future: AI automatically uses Zustand
-```
-
----
-
-### 2. Checkpoint Optimization
-
-Learn which checkpoints user actually reviews:
-
-```yaml
-checkpoint_adaptation:
-  # If user skips checkpoint >80% of time
-  high_skip_rate:
-    action: "suggest_disable"
-    threshold: 0.8
-
-  # If user always reviews carefully
-  careful_review:
-    action: "keep_enabled"
-    indicators:
-      - "review_time > 2min"
-      - "makes_changes"
-
-  current_optimization:
-    after_requirements: "suggest_skip"  # 90% skip rate
-    after_design: "keep"                # Always reviewed
-    after_testing: "suggest_skip"       # 85% skip rate
-    before_deployment: "keep"           # Always reviewed
-```
-
----
-
-### 3. Pattern Following
-
-Automatically follow established codebase patterns:
-
-```yaml
-pattern_adaptation:
-  # Component creation
-  when: "creating_component"
-  learned:
-    location: "src/components/{category}/"
-    structure: "folder-per-component"
-    includes:
-      - "{Name}.tsx"
-      - "{Name}.test.tsx"
-      - "index.ts"
-    style: "functional with memo"
-    props: "interface above component"
-
-  # Service creation
-  when: "creating_service"
-  learned:
-    location: "src/services/"
-    naming: "{name}Service.ts"
-    pattern: "object with methods"
-    includes_types: true
-```
-
----
-
-### 4. Output Calibration
-
-Adjust output based on user preferences:
-
-```yaml
-output_adaptation:
-  verbosity:
-    learned: "moderate"
-    indicators:
-      - "user rarely reads long explanations"
-      - "prefers bullet points"
-      - "wants code more than prose"
-
-  code_style:
-    learned:
-      comments: "minimal"
-      type_annotations: "comprehensive"
-      error_handling: "always_include"
-
-  communication:
-    learned:
-      announce_decisions: true
-      explain_reasoning: "brief"
-      suggest_alternatives: false
-```
-
----
-
-### 5. Decision Prediction
-
-Predict decisions based on history:
-
-```yaml
-decision_prediction:
-  # High confidence predictions
-  high_confidence:
-    - decision: "state_management"
-      prediction: "zustand"
-      confidence: 0.95
-      basis: "100% historical choice"
-
-    - decision: "styling"
-      prediction: "tailwind"
-      confidence: 0.92
-      basis: "used in all features"
-
-  # Medium confidence
-  medium_confidence:
-    - decision: "error_handling"
-      prediction: "error_boundary"
-      confidence: 0.70
-      basis: "used in 70% of components"
-
-  # Apply predictions
-  application:
-    high_confidence: "auto_apply"
-    medium_confidence: "suggest_with_note"
-    low_confidence: "present_options"
-```
-
----
-
-## Adaptation Rules
-
-### Rule 1: Conservative Learning
-
-```yaml
-conservative_learning:
-  # Don't learn from one-off decisions
-  minimum_occurrences: 3
-
-  # Decay old patterns
-  pattern_decay:
-    after_days: 90
-    reduce_confidence_by: 0.1
-
-  # Allow override
-  user_can_override: true
-```
-
-### Rule 2: Context-Aware Application
-
-```yaml
-context_awareness:
-  # Different contexts may have different patterns
-  contexts:
-    - context: "component"
-      patterns: [...]
-    - context: "service"
-      patterns: [...]
-    - context: "api"
-      patterns: [...]
-
-  # Match context before applying
-  matching:
-    use_file_location: true
-    use_task_description: true
-    use_existing_imports: true
-```
-
-### Rule 3: Reversible Adaptations
-
-```yaml
-reversibility:
-  # All adaptations can be undone
-  undo_commands:
-    - "pa:learning-undo last"
-    - "pa:learning-undo [adaptation-id]"
-    - "pa:learning-disable [pattern]"
-
-  # Reset to defaults
-  reset:
-    - "pa:learning-reset"
-    - "Clears all learned patterns"
-    - "Returns to default behavior"
-```
-
----
-
-## Adaptation Feedback Loop
-
-```
-┌─────────────────────────────────────────────────────────────────┐
-│                    Learning Feedback Loop                        │
-├─────────────────────────────────────────────────────────────────┤
-│                                                                 │
-│   User Action → Collect Data → Analyze Pattern → Apply Learning │
-│        ↑                                              │         │
-│        │                                              ↓         │
-│        └──────────── Verify Outcome ←─────────────────┘         │
-│                                                                 │
-│   If outcome = positive → Reinforce pattern                    │
-│   If outcome = negative → Reduce confidence                    │
-│   If outcome = correction → Learn new pattern                  │
-│                                                                 │
-└─────────────────────────────────────────────────────────────────┘
-```
-
----
-
-## Adaptation Dashboard
-
-```
-┌─────────────────────────────────────────────────────────────────┐
-│ Learning Adaptations - Active                                   │
-├─────────────────────────────────────────────────────────────────┤
-│ Auto-Applied Corrections: 12                                    │
-│ ├── State management: Zustand (5 corrections learned)          │
-│ ├── Component style: Functional (3 corrections learned)        │
-│ └── File naming: kebab-case (4 corrections learned)            │
-├─────────────────────────────────────────────────────────────────┤
-│ Checkpoint Optimizations:                                       │
-│ ├── after_requirements: Suggested skip (90% skip rate)         │
-│ └── after_design: Keep (always reviewed)                       │
-├─────────────────────────────────────────────────────────────────┤
-│ Pattern Following: 45 patterns learned                          │
-│ ├── Component patterns: 15                                     │
-│ ├── Service patterns: 10                                       │
-│ ├── Test patterns: 12                                          │
-│ └── Import patterns: 8                                         │
-├─────────────────────────────────────────────────────────────────┤
-│ Decision Predictions:                                           │
-│ ├── High confidence: 8                                         │
-│ ├── Medium confidence: 5                                       │
-│ └── Learning: 3                                                │
-└─────────────────────────────────────────────────────────────────┘
-```
-
----
-
-## Configuration
-
-```yaml
-# proagents.config.yaml
-
-learning:
-  adaptation:
-    # Enable auto-corrections
-    auto_corrections:
-      enabled: true
-      threshold: 3
-
-    # Checkpoint optimization
-    checkpoint_optimization:
-      enabled: true
-      suggest_disable_threshold: 0.8
-
-    # Pattern following
-    pattern_following:
-      enabled: true
-      min_confidence: 0.7
-
-    # Decision prediction
-    decision_prediction:
-      enabled: true
-      auto_apply_threshold: 0.9
-
-    # Safety
-    safety:
-      require_confirmation_for_new: false
-      allow_override: true
-      max_auto_corrections_per_session: 10
-```
-
----
-
-## Commands
-
-| Command | Description |
-|---------|-------------|
-| `pa:adaptation-status` | View current adaptations |
-| `pa:adaptation-disable [id]` | Disable specific adaptation |
-| `pa:adaptation-enable [id]` | Re-enable adaptation |
-| `pa:adaptation-reset` | Reset all adaptations |
-| `pa:adaptation-review` | Review pending adaptations |

package/.proagents/learning/data-collection.md

@@ -1,283 +0,0 @@
-# Learning Data Collection
-
-What data ProAgents collects to improve over time.
-
----
-
-## Overview
-
-ProAgents learns from your development patterns to provide better suggestions and reduce friction over time.
-
----
-
-## Data Categories
-
-### 1. User Preferences
-
-**What's Collected:**
-```yaml
-preferences:
-  # Checkpoint preferences
-  checkpoints:
-    usually_skips: ["after_requirements", "after_testing"]
-    always_reviews: ["after_design", "before_deployment"]
-    skip_rate_by_phase:
-      analysis: 0.2
-      requirements: 0.8
-      design: 0.1
-
-  # Detail level
-  output_preferences:
-    verbosity: "moderate"  # concise | moderate | detailed
-    code_comments: "minimal"
-    documentation_depth: "moderate"
-
-  # Review patterns
-  review_behavior:
-    average_review_time: "5min"
-    usually_accepts_first_suggestion: true
-    common_modifications: ["naming", "structure"]
-```
-
-**How It's Used:**
-- Skip unnecessary checkpoints automatically
-- Adjust output verbosity
-- Pre-modify suggestions based on patterns
-
----
-
-### 2. Correction History
-
-**What's Collected:**
-```yaml
-corrections:
-  - id: "corr_001"
-    timestamp: "2024-01-15T10:30:00Z"
-    original_suggestion: "Use Redux for state management"
-    user_correction: "Use Zustand instead"
-    context: "new_feature"
-    frequency: 3  # How many times this correction happened
-
-  - id: "corr_002"
-    timestamp: "2024-01-16T14:00:00Z"
-    original_suggestion: "Create class component"
-    user_correction: "Use functional component"
-    context: "component_creation"
-    frequency: 5
-```
-
-**How It's Used:**
-- After 3 similar corrections → Auto-apply
-- Prevent repeated mistakes
-- Learn user's preferred approaches
-
----
-
-### 3. Project Patterns
-
-**What's Collected:**
-```yaml
-project_patterns:
-  # File structure
-  structure:
-    components_location: "src/components/"
-    services_location: "src/services/"
-    hooks_location: "src/hooks/"
-    pattern_confidence: 0.95
-
-  # Naming conventions
-  naming:
-    components: "PascalCase"
-    hooks: "useCamelCase"
-    services: "camelCaseService"
-    files:
-      components: "PascalCase.tsx"
-      utilities: "camelCase.ts"
-    pattern_confidence: 0.92
-
-  # Code patterns
-  code_style:
-    component_style: "functional"
-    state_management: "zustand"
-    data_fetching: "react-query"
-    styling: "tailwind"
-    pattern_confidence: 0.98
-
-  # Import patterns
-  imports:
-    style: "named"
-    order: ["react", "third-party", "local", "styles"]
-    use_aliases: true
-    alias_prefix: "@/"
-```
-
-**How It's Used:**
-- Generate code matching existing patterns
-- Place files in correct locations
-- Follow established conventions
-
----
-
-### 4. Decision History
-
-**What's Collected:**
-```yaml
-decisions:
-  - decision_id: "dec_001"
-    type: "architecture"
-    question: "API style selection"
-    chosen: "rest"
-    alternatives: ["graphql", "grpc"]
-    context: "user_service"
-    outcome: "successful"  # successful | needed_revision | failed
-
-  - decision_id: "dec_002"
-    type: "implementation"
-    question: "Error handling approach"
-    chosen: "error-boundary"
-    alternatives: ["try-catch", "result-type"]
-    context: "component"
-    outcome: "successful"
-```
-
-**How It's Used:**
-- Recommend previously successful decisions
-- Avoid approaches that needed revision
-- Build decision confidence scores
-
----
-
-### 5. Time Metrics
-
-**What's Collected:**
-```yaml
-time_metrics:
-  # Phase durations
-  phases:
-    analysis:
-      average: "15min"
-      min: "5min"
-      max: "45min"
-    implementation:
-      average: "2hrs"
-      min: "30min"
-      max: "8hrs"
-
-  # Feature completion
-  features:
-    average_completion: "4hrs"
-    by_complexity:
-      simple: "1hr"
-      medium: "3hrs"
-      complex: "8hrs"
-
-  # Review times
-  review:
-    average_checkpoint_review: "5min"
-    pr_review_time: "30min"
-```
-
-**How It's Used:**
-- Better task estimation
-- Identify bottleneck phases
-- Optimize workflow suggestions
-
----
-
-### 6. Error Patterns
-
-**What's Collected:**
-```yaml
-errors:
-  # Common errors
-  - pattern: "import_path_error"
-    frequency: 12
-    usual_cause: "wrong_alias"
-    usual_fix: "use @/ prefix"
-
-  - pattern: "type_mismatch"
-    frequency: 8
-    usual_cause: "optional_property"
-    usual_fix: "add undefined check"
-
-  # Build failures
-  build_failures:
-    - type: "typescript_error"
-      frequency: 5
-      common_fixes: ["add_types", "fix_imports"]
-
-    - type: "test_failure"
-      frequency: 3
-      common_fixes: ["update_mocks", "fix_assertions"]
-```
-
-**How It's Used:**
-- Proactively prevent common errors
-- Suggest fixes based on history
-- Improve code generation accuracy
-
----
-
-## Storage Structure
-
-```
-.proagents/learning/
-├── global/                      # Cross-project learnings
-│   ├── user-preferences.json
-│   └── common-corrections.json
-│
-└── projects/
-    └── [project-hash]/
-        ├── patterns.json        # Project-specific patterns
-        ├── decisions.json       # Decision history
-        ├── corrections.json     # Corrections in this project
-        ├── metrics.json         # Time and performance metrics
-        └── errors.json          # Error patterns
-```
-
----
-
-## Privacy & Data Control
-
-### What's NOT Collected
-- Source code content
-- Credentials or secrets
-- Personal identifiable information
-- Third-party API keys
-
-### Data Stays Local
-```yaml
-privacy:
-  data_location: "local"  # All data stays on your machine
-  share_anonymized: false  # No cloud sharing by default
-  retention: "project_lifetime"
-```
-
-### User Controls
-```yaml
-learning:
-  # Enable/disable specific tracking
-  track_preferences: true
-  track_patterns: true
-  track_corrections: true
-  track_metrics: true
-  track_errors: true
-
-  # Reset options
-  reset_commands:
-    - "pa:learning-reset all"      # Reset everything
-    - "pa:learning-reset project"  # Reset current project
-    - "pa:learning-reset patterns" # Reset patterns only
-```
-
----
-
-## Commands
-
-| Command | Description |
-|---------|-------------|
-| `pa:learning-status` | Show learning statistics |
-| `pa:learning-patterns` | View learned patterns |
-| `pa:learning-corrections` | View correction history |
-| `pa:learning-reset` | Reset learning data |
-| `pa:learning-export` | Export learning data |