stemedb/applications/aphoria/dogfood/dbpool/RESET-2026-02-09.md

Dogfood Directory Reset - 2026-02-09

Summary

Reset dbpool dogfood directory for next team run after evaluation identified critical documentation gaps.

What Happened

Previous Run (2026-02-09):

• Team followed CHECKLIST.md Day 1
• Fetched all 3 authority source documents ✓
• Created 0 claims (expected 25-30) ✗
• Believed Day 1 was 90% complete (actually 10%)
• Had to "go somewhere else" to learn about flywheel configuration

Root Cause: CHECKLIST.md structured Day 1 as "Information Needed" with checkboxes only for source fetching. Actual deliverable (creating 25-30 claims) was prose without checkboxes, causing team to interpret source fetching as completion.

Evaluation Reports: See eval/ directory for complete analysis

---

Documentation Fixes Applied

1. CHECKLIST.md Day 1 Restructure

• ✅ Changed heading to "Create 25-30 Corpus Claims"
• ✅ Added success criteria at top (verification command)
• ✅ Added estimated time (4-6 hours)
• ✅ Converted claim creation to 27 checkbox items (grouped by category)
• ✅ Added "Now Apply This" practice bridge with 3 practice claims
• ✅ Added step numbers (Step 1, 2, 3, 4)
• ✅ Added explicit completion criteria

2. Flywheel Documentation (NEW)

• ✅ Created docs/flywheel-setup.md with complete configuration guide
• ✅ Updated Day 3 in CHECKLIST.md to reference flywheel setup
• ✅ Added critical section "Configure Flywheel Before Scanning"
• ✅ Updated all scan commands to use --persist flag
• ✅ Updated CLAUDE.md with flywheel references

3. Configuration

• ✅ .aphoria/config.toml already has mode = "persistent"
• ✅ Already has aggregation_enabled = true
• ✅ Full flywheel configuration with comments

---

Files Reset

Removed

src/                     # Placeholder implementation (1 file)
tests/                   # Empty directory
Cargo.toml               # Did not exist
scan-results-*.json      # Did not exist

Moved to eval/

IMPLEMENTATION-SUMMARY.md  # Previous run notes

Preserved

✅ CHECKLIST.md            # UPDATED with fixes
✅ CLAUDE.md               # UPDATED with flywheel refs
✅ plan.md                 # Original plan
✅ README.md               # NEW reset guide
✅ docs/
   ✅ claim-extraction-example.md  # Original
   ✅ flywheel-setup.md            # NEW
   ✅ sources/                     # All 3 source docs preserved
      ✅ hikaricp-config.md
      ✅ owasp-credentials.md
      ✅ postgresql-pooling.md
✅ .aphoria/config.toml    # Flywheel configured
✅ .claude/                # Claude Code config
✅ scripts/                # Pre-flight validator
✅ eval/                   # Previous run analysis

---

Directory Structure After Reset

dbpool/
├── README.md                    # NEW: Reset guide
├── CHECKLIST.md                 # UPDATED: Fixed Day 1
├── CLAUDE.md                    # UPDATED: Flywheel refs
├── plan.md                      # Original
├── RESET-2026-02-09.md          # This file
├── .aphoria/
│   ├── config.toml              # Flywheel configured
│   └── agent.key                # Signing key
├── .claude/
│   └── settings.local.json      # Claude settings
├── docs/
│   ├── claim-extraction-example.md   # Original
│   ├── flywheel-setup.md             # NEW
│   └── sources/
│       ├── hikaricp-config.md        # Preserved
│       ├── owasp-credentials.md      # Preserved
│       └── postgresql-pooling.md     # Preserved
├── eval/
│   ├── EVALUATION-REPORT-2026-02-09.md
│   ├── gap-analysis-2026-02-09.md
│   ├── implementation-review-2026-02-09.md
│   ├── progress-log-2026-02-09.md
│   └── IMPLEMENTATION-SUMMARY.md     # Moved from root
└── scripts/
    └── validate-setup.sh              # Pre-flight validator

MISSING (will be created during exercise):
- src/        # Day 2
- tests/      # Day 2
- Cargo.toml  # Day 2

---

Verification

Pre-Flight Check

./scripts/validate-setup.sh
# Should pass all checks

Documentation Complete

# Verify all docs exist
ls -1 docs/
# Should show:
#   claim-extraction-example.md
#   flywheel-setup.md
#   sources/

# Verify Day 1 has clear deliverable
head -120 CHECKLIST.md | grep "Create 25-30"
# Should show: "## Day 1: Create 25-30 Corpus Claims"

# Count claim checkboxes
grep -c "- \[ \].*dbpool/" CHECKLIST.md
# Should show: 27 (or more with verification steps)

Configuration Verified

# Check flywheel mode
grep "mode.*persistent" .aphoria/config.toml
# Output: mode = "persistent"  # Required for pattern aggregation

# Check aggregation enabled
grep "aggregation_enabled" .aphoria/config.toml
# Output: aggregation_enabled = true  # Default: true (CRITICAL for flywheel)

Source Documents Preserved

ls -1 docs/sources/
# Should show:
#   hikaricp-config.md
#   owasp-credentials.md
#   postgresql-pooling.md

# These were already fetched by previous team
# Next team can skip source fetching (already done)

---

Expected Outcomes

Previous Run

• Completion rate: 10% (0/27 claims created)
• Team confusion: Thought Day 1 was 90% complete
• Missing documentation: Had to find flywheel info elsewhere

Next Run (Expected)

• Completion rate: 85-90% (25-27 claims created)
• Clear deliverable: 27 checkbox items impossible to miss
• Complete documentation: Flywheel guide included
• Practice bridge: 3 practice claims before full set
• Explicit verification: Success criteria at top

---

Next Team Instructions

1. Run pre-flight validation:

   ./scripts/validate-setup.sh
   

2. Read the reset guide:

   cat README.md
   

3. Read Day 1 checklist:

   cat CHECKLIST.md | head -300
   

4. Start with claim extraction example:

   cat docs/claim-extraction-example.md
   

5. Begin Day 1:

   • Follow CHECKLIST.md step by step
   • Complete all 27 claim checkboxes
   • Verify with success criteria command
   • Should take 4-6 hours

6. Before Day 3:

   • Read docs/flywheel-setup.md
   • Verify config has mode = "persistent"

---

Files Modified

File	Status	Changes
CHECKLIST.md	UPDATED	Day 1 restructure, 27 checkboxes, practice bridge, step numbers
CLAUDE.md	UPDATED	Added flywheel references and commands
docs/flywheel-setup.md	NEW	Complete flywheel configuration guide
README.md	NEW	Reset guide and quick start
RESET-2026-02-09.md	NEW	This documentation
.aphoria/config.toml	UNCHANGED	Already configured correctly
docs/sources/*.md	UNCHANGED	Preserved from previous run
src/	REMOVED	Placeholder implementation deleted
tests/	REMOVED	Empty directory deleted
IMPLEMENTATION-SUMMARY.md	MOVED	Moved to eval/

---

Success Metrics

After reset, next team should achieve:

• ✅ 25-30 claims created (vs 0 in previous run)
• ✅ Clear understanding of deliverable
• ✅ No "where do I find this?" questions
• ✅ Smooth Day 1 → Day 2 transition
• ✅ Complete flywheel understanding before Day 3

Target: 85-90% cold-start success rate

---

Reset Date: 2026-02-09 Reset By: Claude Code (based on team evaluation) Evaluation Reports: See eval/ directory Ready For: Next team run with improved documentation