feat: Add learning capture and pre-flight check system for THEALGORITHM

Packs/pai-algorithm-skill/src/skills/THEALGORITHM/Phases/Learn.md

@@ -1,19 +1,71 @@
 # LEARN Phase
 
-**Purpose:** OUTPUT results for user to evaluate. User rates outputs for the memory system.
+**Purpose:** OUTPUT results for user to evaluate AND capture learnings for the memory system.
 
-**ISC Mutation:** ARCHIVE completed ISC, output results
+**ISC Mutation:** ARCHIVE completed ISC, output results, capture learnings
 
-**Gate Question:** Have I provided everything the user needs to evaluate the work?
+**Gate Question:** Have I provided everything the user needs to evaluate the work AND captured learnings?
 
 **CRITICAL:** NO SELF-RATING. The assistant does NOT rate its own work. User rates outputs for the memory system.
 
 ## What Happens
 
 1. **Output final ISC** - Show completed state with all capabilities and statuses
 2. **Present deliverables** - Show what was produced
-3. **Document capability usage** - What capabilities were used and how they performed
-4. **Archive or iterate** - Complete if successful, iterate if blocked
+3. **Capture learnings** - Extract and persist insights (AUTOMATED)
+4. **Record signals** - Track patterns, failures, loopbacks (AUTOMATED)
+5. **Archive or iterate** - Complete if successful, iterate if blocked
+
+## Learning Capture (NEW - AUTOMATED)
+
+The LEARN phase now automatically captures learnings using the new tools:
+
+### Quick Commands
+
+```bash
+# Full learning capture (RECOMMENDED - does everything)
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/ISCManager.ts learn
+
+# Or individual tools:
+
+# 1. Capture learnings from ISC
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/LearningCapture.ts capture
+
+# 2. Record task completion
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/SignalCapture.ts complete -w "task name" -i 1 --success
+
+# 3. Analyze patterns
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/SignalCapture.ts analyze
+
+# 4. Complete work item
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/WorkItemManager.ts complete --success -s "summary"
+```
+
+### What Gets Captured
+
+| Type | Location | Description |
+|------|----------|-------------|
+| **Capability Performance** | `MEMORY/Learning/EXECUTE/` | How well each capability performed |
+| **Estimation Accuracy** | `MEMORY/Learning/PLAN/` | Was effort level appropriate? |
+| **Patterns** | `MEMORY/Learning/ALGORITHM/` | Blocked/adjusted row patterns |
+| **Session Summary** | `MEMORY/Learning/sessions/` | Markdown summary of the run |
+| **Signals** | `MEMORY/Signals/patterns.jsonl` | Aggregated patterns for analysis |
+
+### Signal Tracking
+
+```bash
+# Record a failure during VERIFY
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/SignalCapture.ts failure \
+  -w "Add auth" -c "tests pass" -e "all green" -o "3 failing"
+
+# Record a loopback
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/SignalCapture.ts loopback \
+  -w "Add auth" --from VERIFY --to THINK -r "unclear requirements"
+
+# Record user rating (after user feedback)
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/SignalCapture.ts rating \
+  -w "Add auth" -s 4 -f "good work"
+```
 
 ## NO Self-Rating
 
@@ -27,7 +79,8 @@
 - Present the work clearly
 - Show what was accomplished
 - Document what capabilities were used
-- Let user evaluate and rate for memory
+- Capture learnings automatically
+- Let user evaluate and rate
 
 ## Output Format
 
@@ -57,122 +110,120 @@ Present results clearly for user to evaluate:
 | execution.engineer | Success | Parallel tasks efficient |
 ...
 
+## Learnings Captured
+
+✅ [X] learnings automatically captured to MEMORY/Learning/
+✅ [X] patterns analyzed and saved to MEMORY/Signals/
+✅ [X] session summary created
+
 ## For Memory System
 
-User: Please rate this work for the memory system.
+User: Please rate this work (1-5) for the memory system.
 ```
 
-## Learning Categories
+## Full LEARN Phase Workflow
 
-| Category | What to Document | For |
-|----------|------------------|-----|
-| **Capability Usage** | Which capabilities were effective? | Future task planning |
-| **Process** | What phases were hard? Easy? | Algorithm improvement |
-| **Estimation** | Was effort level accurate? | Calibration |
+```bash
+# 1. Transition to LEARN phase
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/AlgorithmDisplay.ts phase LEARN
 
-## Review the Evolution
+# 2. Show final ISC
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/ISCManager.ts show -o markdown
 
-```bash
-# View the full evolution log
-bun run ISCManager.ts log
+# 3. Capture all learnings (AUTOMATED)
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/ISCManager.ts learn
+
+# 4. Present to user for rating
 
-# View final ISC state
-bun run ISCManager.ts show -o markdown
+# 5. After user rates (if they do):
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/SignalCapture.ts rating -w "task" -s [1-5] -f "feedback"
 
-# Get summary statistics
-bun run ISCManager.ts summary
+# 6. Archive ISC (also captures learnings if not done)
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/ISCManager.ts clear
 ```
 
-## Document Capability Usage
+## Memory Locations
 
-For each capability used, note:
-- **What capability:** e.g., `research.perplexity`, `execution.engineer`
-- **What it did:** Brief description of how it was used
-- **Effectiveness:** Did it work well for this task?
+| Path | Purpose |
+|------|---------|
+| `$PAI_DIR/MEMORY/Learning/ALGORITHM/` | Algorithm-specific learnings |
+| `$PAI_DIR/MEMORY/Learning/EXECUTE/` | Capability performance |
+| `$PAI_DIR/MEMORY/Learning/PLAN/` | Estimation accuracy |
+| `$PAI_DIR/MEMORY/Learning/sessions/` | Session summaries (markdown) |
+| `$PAI_DIR/MEMORY/Signals/patterns.jsonl` | Aggregated patterns |
+| `$PAI_DIR/MEMORY/Signals/failures.jsonl` | Failure records |
+| `$PAI_DIR/MEMORY/Signals/loopbacks.jsonl` | Loopback records |
+| `$PAI_DIR/MEMORY/Signals/ratings.jsonl` | User ratings |
+| `$PAI_DIR/MEMORY/State/algorithm-stats.json` | Overall statistics |
+| `$PAI_DIR/MEMORY/State/algorithm-streak.json` | Success streak |
 
-This helps improve future capability selection.
+## Viewing Learnings
+
+```bash
+# List recent learnings
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/LearningCapture.ts list -l 20
+
+# Show learning statistics
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/LearningCapture.ts stats
+
+# Show signal report
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/SignalCapture.ts report
+
+# Show algorithm stats
+bun run $PAI_DIR/skills/THEALGORITHM/Tools/SignalCapture.ts stats
+```
+
+## Learning Categories
+
+| Category | What to Document | For |
+|----------|------------------|-----|
+| **Capability Usage** | Which capabilities were effective? | Future task planning |
+| **Process** | What phases were hard? Easy? | Algorithm improvement |
+| **Estimation** | Was effort level accurate? | Calibration |
+| **Patterns** | What patterns emerge? | Long-term optimization |
 
 ## Iteration Decision
 
 | Outcome | Action |
 |---------|--------|
-| All rows PASS/ADJUSTED | Archive ISC, output for user |
+| All rows PASS/ADJUSTED | Capture learnings, archive ISC, output for user |
 | Some rows BLOCKED | Iterate - return to appropriate phase |
 | Scope fundamentally wrong | Iterate from OBSERVE |
 
 **Iteration flow:**
 ```
 BLOCKED in verification
     ↓
+Record failure signal
+    ↓
 Analyze blockers
     ↓
 If implementation issue → Return to EXECUTE
 If plan issue → Return to PLAN
 If requirements issue → Return to THINK
 If scope issue → Return to OBSERVE
     ↓
+Record loopback signal
+    ↓
 Increment iteration counter
     ↓
 Re-run from that phase
 ```
 
-## Archive Completed Work
-
-```bash
-# Archive the ISC (saves to archive, clears current)
-bun run ISCManager.ts clear
-
-# This creates: $PAI_DIR/MEMORY/Work/archive-{timestamp}.json
-```
-
-## Example Output
-
-**Request:** "Add logout button to navbar"
-**Effort:** STANDARD | **Iterations:** 1
-
-**ISC Final State:**
-- 7 rows total
-- 6 PASS, 1 ADJUSTED
-- 2 parallel execution groups
-
-**Capabilities Used:**
-| Capability | Result | Notes |
-|------------|--------|-------|
-| execution.engineer×2 | Success | Parallel UI tasks |
-| execution.qa_tester | Success | Caught mobile issue |
-| verification.browser | Success | Visual verification |
-
-**Deliverables:**
-- Logout button component in `components/LogoutButton.tsx`
-- Updated navbar with logout integration
-- Tests passing, browser-verified
-
-**For Memory System:**
-User: Please rate this work for the memory system.
-
-## Commands
-
-```bash
-# Update phase
-bun run ISCManager.ts phase -p LEARN
-
-# Start new iteration (if needed)
-bun run ISCManager.ts iterate
-
-# Archive completed ISC
-bun run ISCManager.ts clear
-```
-
 ## Exit Criteria
 
 **If successful:**
 - Final ISC displayed with capabilities
 - Deliverables presented
-- Capability usage documented
+- ✅ Learnings captured to MEMORY/Learning/
+- ✅ Signals recorded to MEMORY/Signals/
+- ✅ Session summary created
 - Ready for user's evaluation
 
 **If iterating:**
 - Blockers analyzed
+- Failure signals recorded
+- Loopback signals recorded
 - Return phase identified
 - Iteration counter incremented
 - Ready to re-enter earlier phase