Signal Providers

BaseCred consumes existing reputation systems as signals. It does not replace them.

External Reputation Signals

Provider	Primary Authority
Ethos	Long-term trust & credibility
Talent Protocol	Ability & skill (builder/creator)
Neynar	Social behavior & spam risk

These systems remain the source of truth for their respective domains.

BaseCred Stance on Scores

• BaseCred does not generate a unified or proprietary score
• BaseCred does not persist score history
• BaseCred does not compete with signal providers

Scores are treated as raw inputs, normalized only for semantic comparison.

Ephemeral Scores

BaseCred does not store score history.

• Scores are derived on demand from current signals
• Scores are ephemeral and recomputable
• Scores are not treated as state

This reduces data liability, avoids stale reputation, and keeps the system deterministic.

Normalized Signals

interface NormalizedSignals {
  trust: Tier; // from Ethos
  socialTrust: Tier; // from Neynar
  builder: Capability; // from Talent
  creator: Capability; // from Talent
  recencyDays: number; // freshness
  spamRisk: Tier; // from Neynar
  signalCoverage: number; // 0-1
}

Normalized signals exist only at decision time and are not persisted.

Normalization Logic & Thresholds

Determined by src/engine/normalizers/.

Signal	Source	Logic / Thresholds
trust	Ethos	Based on credibility_score: • VERY_HIGH ≥ 40 • HIGH ≥ 20 • NEUTRAL ≥ 0 • LOW ≥ -20 • Else VERY_LOW
socialTrust	Neynar	Based on farcaster_user_score (0-1): • VERY_HIGH ≥ 0.9 • HIGH ≥ 0.7 • NEUTRAL ≥ 0.4 • LOW ≥ 0.2 • Else VERY_LOW
spamRisk	Neynar	Inverse of farcaster_user_score: • VERY_LOW (Safe) ≥ 0.8 • LOW ≥ 0.6 • NEUTRAL ≥ 0.4 • HIGH ≥ 0.2 • Else VERY_HIGH (Risky)
builder	Talent Protocol	Based on builder.score: • EXPERT ≥ 80 • ADVANCED ≥ 50 • INTERMEDIATE ≥ 20 • Else NONE
creator	Talent Protocol	Based on creator.score: • EXPERT ≥ 80 • ADVANCED ≥ 50 • INTERMEDIATE ≥ 20 • Else NONE