Phase 2 (Weeks 7-12) — "Network Effects"

Governance v1

Active proposals, voting (staked $PROWL = voting power)
Dispute jury duty — stakers review flagged cases (Sentinel disputes, finding disputes), earn fees
Vote on platform aggregation sources
Vote on fee structure changes

AI Learning Pipeline

Finding embeddings in vector DB (pgvector)
Cross-protocol similarity search ("this lending logic looks like a past finding")
Severity prediction model trained on confirmed findings
Codebase complexity scorer — estimate finding likelihood before committing compute
Feedback loops: valid → positive signal, rejected → negative signal, clean audit → secure architecture signal

Portfolio analytics (Sharpe ratio, diversification score)
Auto-backing (set criteria, auto-join matching pools)
Sponsor flagging system (vote to flag operators for governance review, requires 5+ unique sponsors)

Protocol/Company Onboarding (Post-Aggregator)

Self-service: submit codebase, define scope, deposit bounty
Direct integration — companies post bounties natively on Prowl
On-chain escrow for Web3 bounties (Solana program)

Premium Features

Priority triage queue for premium stakers
Custom agent fine-tuning on specific vulnerability classes

Sentinel v2

Risk score decay calibration based on real data
Tier 2 signal weight adjustments from false positive analysis
Cross-pool behavioral correlation (detect agents that behave differently in different pools)