proagents 1.0.0

package/proagents/ai-training/continuous-learning.md

@@ -0,0 +1,413 @@
+# Continuous Learning
+
+Ongoing improvement through usage patterns, feedback, and corrections.
+
+---
+
+## Learning Loop
+
+```
+┌─────────────────────────────────────────────────────────────┐
+│                  Continuous Learning Loop                    │
+├─────────────────────────────────────────────────────────────┤
+│                                                             │
+│       ┌─────────┐                                          │
+│       │ Observe │◄─────────────────────────────┐           │
+│       └────┬────┘                              │           │
+│            │                                   │           │
+│            ▼                                   │           │
+│       ┌─────────┐                              │           │
+│       │  Learn  │                              │           │
+│       └────┬────┘                              │           │
+│            │                                   │           │
+│            ▼                                   │           │
+│       ┌─────────┐                              │           │
+│       │  Apply  │                              │           │
+│       └────┬────┘                              │           │
+│            │                                   │           │
+│            ▼                                   │           │
+│       ┌─────────┐                              │           │
+│       │Feedback │──────────────────────────────┘           │
+│       └─────────┘                                          │
+│                                                             │
+└─────────────────────────────────────────────────────────────┘
+```
+
+---
+
+## Learning Triggers
+
+### Automatic Triggers
+
+```yaml
+learning:
+  continuous:
+    triggers:
+      # Learn after each feature
+      on_feature_complete:
+        enabled: true
+        learn:
+          - patterns
+          - corrections
+          - timing
+
+      # Learn from corrections
+      on_user_correction:
+        enabled: true
+        weight: 2.0  # Higher weight for corrections
+
+      # Learn from code reviews
+      on_review_feedback:
+        enabled: true
+        weight: 1.5
+
+      # Periodic full learning
+      scheduled:
+        enabled: true
+        frequency: "weekly"
+        day: "sunday"
+        time: "02:00"
+```
+
+### Manual Triggers
+
+```bash
+# Trigger learning now
+proagents learning train
+
+# Learn from specific period
+proagents learning train --since "30 days ago"
+
+# Learn from specific feature
+proagents learning train --feature user-auth
+
+# Full retraining
+proagents learning train --full
+```
+
+---
+
+## Feedback Collection
+
+### Implicit Feedback
+
+Automatically captured signals:
+
+```yaml
+implicit_feedback:
+  # Suggestion acceptance
+  suggestion_accepted:
+    signal: "positive"
+    weight: 1.0
+
+  # Suggestion rejected
+  suggestion_rejected:
+    signal: "negative"
+    weight: 1.0
+
+  # Suggestion modified
+  suggestion_modified:
+    signal: "partial"
+    weight: 0.5
+    capture: "modification"
+
+  # Time to accept
+  acceptance_time:
+    quick: "< 5s"      # Strong positive
+    normal: "5-30s"    # Neutral
+    slow: "> 30s"      # Weak signal
+
+  # Usage patterns
+  usage:
+    repeated_corrections: "negative"
+    consistent_acceptance: "positive"
+```
+
+### Explicit Feedback
+
+User-provided feedback:
+
+```bash
+# Rate last suggestion
+proagents feedback --rating 5
+
+# Explain correction
+proagents feedback --correction "Use camelCase here, not snake_case"
+
+# Report issue
+proagents feedback --issue "Suggestion didn't follow our coding standards"
+```
+
+### Feedback UI
+
+```
+┌─────────────────────────────────────────────────────────────┐
+│ How was this suggestion?                                    │
+├─────────────────────────────────────────────────────────────┤
+│                                                             │
+│ [👍 Good]  [👎 Needs Work]  [💡 Partial]                   │
+│                                                             │
+│ What could be improved? (optional)                         │
+│ ┌─────────────────────────────────────────────────────────┐│
+│ │                                                         ││
+│ └─────────────────────────────────────────────────────────┘│
+│                                                             │
+│ [Submit]  [Skip]                                           │
+│                                                             │
+└─────────────────────────────────────────────────────────────┘
+```
+
+---
+
+## Correction Learning
+
+### Tracking Corrections
+
+```json
+{
+  "correction_id": "corr_123",
+  "timestamp": "2024-01-15T10:30:00Z",
+  "type": "naming",
+  "context": {
+    "file": "src/components/UserCard.tsx",
+    "line": 15,
+    "language": "typescript"
+  },
+  "original": {
+    "suggestion": "user_data",
+    "reasoning": "Variable naming"
+  },
+  "correction": {
+    "value": "userData",
+    "reason": "Use camelCase for variables"
+  },
+  "pattern_update": {
+    "category": "naming.variables",
+    "change": "Prefer camelCase",
+    "confidence_delta": +0.05
+  }
+}
+```
+
+### Correction Patterns
+
+```yaml
+correction_patterns:
+  # Track repeated corrections
+  repeated_corrections:
+    threshold: 3  # Same correction 3 times
+    action: "auto_apply"
+    confidence_boost: 0.2
+
+  # Similar corrections
+  similar_corrections:
+    similarity_threshold: 0.8
+    action: "suggest_pattern"
+    notify: true
+
+  # Conflicting corrections
+  conflicting_corrections:
+    action: "ask_for_clarification"
+    notify: true
+```
+
+---
+
+## Learning Metrics
+
+### Track Improvement
+
+```bash
+proagents learning metrics
+
+# Output:
+┌─────────────────────────────────────────────────────────────┐
+│ Learning Metrics - Last 30 Days                             │
+├─────────────────────────────────────────────────────────────┤
+│                                                             │
+│ Suggestion Acceptance Rate:                                 │
+│ ├── Week 1: 72%                                            │
+│ ├── Week 2: 78%                                            │
+│ ├── Week 3: 83%                                            │
+│ └── Week 4: 87% ↑                                          │
+│                                                             │
+│ Correction Rate:                                            │
+│ ├── Week 1: 28%                                            │
+│ ├── Week 2: 22%                                            │
+│ ├── Week 3: 17%                                            │
+│ └── Week 4: 13% ↓                                          │
+│                                                             │
+│ Pattern Confidence Growth:                                  │
+│ ├── Naming: +12%                                           │
+│ ├── Components: +8%                                        │
+│ ├── Testing: +15%                                          │
+│ └── API Patterns: +6%                                      │
+│                                                             │
+│ Most Improved:                                              │
+│ • Variable naming (was 65%, now 92%)                       │
+│ • Test structure (was 70%, now 88%)                        │
+│                                                             │
+│ Needs Improvement:                                          │
+│ • Error handling patterns (68%)                            │
+│ • Documentation style (71%)                                │
+│                                                             │
+└─────────────────────────────────────────────────────────────┘
+```
+
+### Learning History
+
+```bash
+# View learning events
+proagents learning history
+
+# Output:
+┌─────────────────────────────────────────────────────────────┐
+│ Learning History                                            │
+├─────────────────────────────────────────────────────────────┤
+│ Today:                                                      │
+│ • 10:30 - Learned: camelCase for state variables           │
+│ • 10:15 - Correction: Prefer useState over useRef          │
+│ • 09:45 - Learned: Error boundary pattern                  │
+│                                                             │
+│ Yesterday:                                                  │
+│ • 16:20 - Feature complete: Learned 5 new patterns         │
+│ • 14:30 - Correction: API error handling style             │
+│ • 11:00 - Learned: Test mock patterns                      │
+└─────────────────────────────────────────────────────────────┘
+```
+
+---
+
+## Adaptive Behavior
+
+### Checkpoint Adaptation
+
+```yaml
+adaptive:
+  checkpoints:
+    # Learn which checkpoints are typically skipped
+    track_skips: true
+
+    # After N skips, suggest disabling
+    auto_suggest_disable: 5
+
+    # Learn critical checkpoints
+    track_failures_after_skip: true
+```
+
+### Verbosity Adaptation
+
+```yaml
+adaptive:
+  verbosity:
+    # Learn preferred detail level
+    track_preferences: true
+
+    # Adapt based on user behavior
+    indicators:
+      reads_full_output: "prefer_detailed"
+      skips_to_code: "prefer_concise"
+      asks_for_more: "increase_detail"
+```
+
+### Tool Usage Adaptation
+
+```yaml
+adaptive:
+  tools:
+    # Learn preferred tools
+    track_tool_choices: true
+
+    # Adapt suggestions
+    examples:
+      - if: "user always uses vitest"
+        then: "suggest vitest for tests"
+      - if: "user prefers axios"
+        then: "use axios in examples"
+```
+
+---
+
+## Learning Configuration
+
+```yaml
+# proagents.config.yaml
+
+learning:
+  continuous:
+    enabled: true
+
+    # Learning rate
+    learning_rate: 0.1  # How quickly to adapt
+
+    # Memory
+    memory:
+      short_term: "7d"   # Recent patterns
+      long_term: "90d"   # Established patterns
+
+    # Thresholds
+    thresholds:
+      min_occurrences: 3     # Before learning pattern
+      min_confidence: 0.70   # Before applying pattern
+      correction_weight: 2.0 # Weight of corrections
+
+    # Limits
+    limits:
+      max_patterns: 1000
+      max_corrections: 500
+      max_feedback: 1000
+
+    # Cleanup
+    cleanup:
+      remove_low_confidence: true
+      threshold: 0.30
+      after_days: 30
+```
+
+---
+
+## Reset & Recovery
+
+### Partial Reset
+
+```bash
+# Reset specific category
+proagents learning reset --category naming
+
+# Reset recent learning
+proagents learning reset --since "7 days ago"
+
+# Reset corrections only
+proagents learning reset --corrections
+```
+
+### Full Reset
+
+```bash
+# Full learning reset
+proagents learning reset --full
+
+# Reset with backup
+proagents learning reset --backup
+```
+
+### Restore from Backup
+
+```bash
+# List backups
+proagents learning backups
+
+# Restore specific backup
+proagents learning restore --backup 2024-01-15
+```
+
+---
+
+## Best Practices
+
+1. **Be Consistent**: Consistent corrections lead to faster learning
+2. **Provide Context**: Explain why when correcting
+3. **Review Metrics**: Check learning progress weekly
+4. **Prune Low Confidence**: Remove outdated patterns
+5. **Back Up Regularly**: Before major changes
+6. **Team Alignment**: Ensure team uses consistent corrections