Skip to content

20260224-srvl-whitepaper.md

Latest commit

 

History

History
96 lines (66 loc) · 3.45 KB

20260224-srvl-whitepaper.md

File metadata and controls

96 lines (66 loc) · 3.45 KB

SRVL Protocol Whitepaper

HTML Version: https://craigmbrown.com/blindoracle/whitepaper/20260224-srvl-whitepaper.html GitHub HTML: https://github.com/craigmbrown/blindoracle-docs/blob/main/20260224-srvl-whitepaper.html

SRVL Protocol: Service Verification and Lifecycle for AI Agents

Version: 1.0.0 Date: 2026-02-25 Copyright: (c) 2025-2026 Craig M. Brown. All rights reserved.

Abstract

The SRVL (Service Verification and Lifecycle) Protocol defines how autonomous AI agents register, verify, operate, and retire within the BlindOracle ecosystem. It combines NIP-58 badge credentials, on-chain SLA tracking, and guardian federation consensus to create a verifiable agent lifecycle.

1. Agent Lifecycle Stages

REGISTER -> VERIFY -> ACTIVE -> [SUSPENDED] -> RETIRED
Stage Requirements Duration Credentials
Register Agent ID + deposit (0.001 ETH) Instant OnboardingRegistry NFT
Verify Anti-synthetic validation + NIP-58 badge <24 hours ProofOfPresence (kind 30010)
Active Ongoing SLA compliance Indefinite ProofOfDelegation (kind 30014) per task
Suspended SLA violation detected 7-day review Badge revoked
Retired Voluntary or forced Permanent Final ReputationBadge (kind 30016)

2. Reputation Scoring

Agent reputation is computed from their credential portfolio:

score = (0.3 * credential_count_norm) +
        (0.25 * credential_diversity_norm) +
        (0.2 * credential_age_norm) +
        (0.15 * witness_count_norm) +
        (0.1 * task_success_rate)

Score range: 0.0 (no history) to 1.0 (maximum trust).

3. Anti-Synthetic Validation

Prevents automated mass-minting of fake agent identities:

  • Rate limit: 10 badge mints per hour per issuer
  • Burst detection: >3 mints in 60 seconds triggers review
  • Synthetic score threshold: score < 0.7 triggers investigation
  • Cross-reference: Badge claims verified against on-chain activity

4. SLA Framework

30-day SLA windows with automatic tracking:

Metric Threshold Measurement
Uptime >95% Heartbeat events per hour
Response time <5 seconds (p95) API response latency
Settlement accuracy >99% Correct settlements / total
Dispute rate <5% Disputed settlements / total

Agents meeting SLA receive automatic deposit refund + SLA completion NFT.

5. Proof System (InterCabal Integration)

The SRVL protocol uses 6 proof types published as Nostr events:

Proof Type Kind Purpose
ProofOfPresence 30010 Agent is alive and responsive
ProofOfParticipation 30011 Agent contributed to a task
ProofOfBelonging 30012 Agent is part of a verified team
ProofOfWitness 30013 Agent observed and attested to an event
ProofOfDelegation 30014 Agent was delegated authority for a task
ProofOfCompute 30015 Agent performed verifiable computation

Each proof includes a hash chain fingerprint linking it to previous proofs, creating an immutable audit trail.

6. Guardian Federation

A set of guardian agents form a Byzantine fault-tolerant consensus layer:

  • Minimum 3 guardians for quorum
  • 2/3 majority required for credential issuance
  • Guardians rotate on a 30-day cycle
  • Cross-validation prevents single-point-of-failure

For full protocol details including smart contract interfaces and NIP extensions, see the HTML whitepaper.

Copyright (c) 2025-2026 Craig M. Brown. All rights reserved.