Phoenix Task
    |
    v
Is this about running truth cases?
    |
    +-- YES --> phoenix-truth-case-orchestrator
    |           Command: /phoenix-truth
    |
    +-- NO --> Continue
    |
Is this about numeric precision or floating-point issues?
    |
    +-- YES --> phoenix-precision-guard
    |           (parseFloat → Decimal.js, tolerance issues)
    |
    +-- NO --> Continue
    |
Is this about waterfall/carry/clawback calculations?
    |
    +-- YES --> phoenix-waterfall-ledger-semantics
    |           Agent: waterfall-specialist
    |
    +-- NO --> Continue
    |
Is this about XIRR or fee calculations?
    |
    +-- YES --> phoenix-xirr-fees-validator
    |           Agent: xirr-fees-validator
    |
    +-- NO --> Continue
    |
Is this about capital allocation or exit recycling?
    |
    +-- YES --> phoenix-capital-exit-investigator
    |           Agent: phoenix-capital-allocation-analyst
    |
    +-- NO --> Continue
    |
Is this about documentation alignment (JSDoc, calculations.md)?
    |
    +-- YES --> phoenix-docs-sync
    |           Agent: phoenix-docs-scribe
    |
    +-- NO --> Continue
    |
Is this about Monte Carlo, graduation modeling, or MOIC?
    |
    +-- YES --> phoenix-advanced-forecasting
    |           Agent: phoenix-probabilistic-engineer
    |           Command: /phoenix-phase2
    |
    +-- NO --> Continue
    |
Is this about reserve sizing or follow-on allocation?
    |
    +-- YES --> phoenix-reserves-optimizer
    |           Agent: phoenix-reserves-optimizer
    |
    +-- NO --> Continue
    |
Is this about UI/reporting/branding?
    |
    +-- YES --> phoenix-brand-reporting
    |           Agent: phoenix-brand-reporting-stylist
    |
    +-- NO --> Use general phoenix-truth-case-orchestrator

1. /phoenix-truth (run suite)
    |
    v
2. Classify failures:
   - Precision drift → phoenix-precision-guard
   - Waterfall semantics → phoenix-waterfall-ledger-semantics
   - XIRR/fees parity → phoenix-xirr-fees-validator
   - Allocation issues → phoenix-capital-exit-investigator
    |
    v
3. Fix with specialist skill
    |
    v
4. /phoenix-truth (verify fix)

1. phoenix-truth-case-orchestrator
   (create truth cases FIRST)
    |
    v
2. Implement calculation
    |
    v
3. Run skill for domain:
   - Waterfall → phoenix-waterfall-ledger-semantics
   - XIRR → phoenix-xirr-fees-validator
   - Reserves → phoenix-reserves-optimizer
    |
    v
4. phoenix-docs-sync
   (align documentation)
    |
    v
5. /phoenix-truth (full validation)

Phase 0 Gate: 100% truth cases passing
    ↓
Phase 1A Gate: Precision issues resolved
    ↓
Phase 1B Gate: Core calculations Excel-parity
    ↓
Phase 2 Gate: Monte Carlo validation stable
    ↓
Phase 3: Brand consistency applied