Diagnostics surfaced 8 stacked failure modes in the validation infrastructure: 3 harnesses producing 3 different Sharpes per strategy, gate not firing in backtest (wall-clock cooldown bug), stale checkpoints masking harness upgrade, phase3 engine bypasses gate entirely, t1_trend_pullback live with no on-disk artifact, 1619% live notional sized on gate-blind PFs. The strategies are real — the measurement layer was lying. 6 phases below restore trust to the published numbers, the live portfolio, and the system. Tap YES / NO / MODIFY. Several tasks are explicit DECISIONS — please use MODIFY to answer them.
Comparable products charge $300–2 000/yr. Recommended tiers:
| Tier | Price | Limits |
|---|---|---|
| Scout | Free | 1 profile · 5 matches/mo · email-only · no briefs |
| Hunter | $79/mo · $790/yr | Unlimited matches · daily/weekly digest · Telegram + email · 1 brief/mo |
| Predator | $199/mo · $1 990/yr | Hunter + unlimited briefs · Slack · 3 user seats · priority support |
Plus one-off $99 "first brief" for non-subscribers — converts cold guide readers.
Once submitted, Claude reads your responses and starts executing. You can still come back and edit, but rerunning the build manually after edits is on you.