Home

Integrity Protocol Wiki

Compiled knowledge base for the Integrity Protocol monorepo — a trust/ compliance layer for AI agents on Base L2. This page is the map; every section below links to the real page. Governance/conventions: WIKI_SCHEMA.md (page format), WIKI_INDEX.md (the full catalog with one-line descriptions — the canonical index this page summarizes), WIKI_LOG.md (chronological history, append-only). Cross-package decisions live in ../INTERFACE_CONTRACT.md; how this wiki gets kept in sync with the code is ../../.agents/AGENTS.md.

Start here if you're new: The Four Foundational Primitives (the concepts), then Agent Primitives (the 7 per-agent contracts every other page assumes you understand) and Persistent Memory (the continuity primitive that gates registration), then AIS (the trust score) and Telemetry Ingestion Pipeline (how agent behavior becomes that score).

Current cross-repository audit — 2026-08-06

The four-repository implementation plan and audit ledger are maintained outside the generated wiki at /home/xibalba/Documents/INTEGRITY — Cross-Repository Audit and Implementation Plan.md. Repository-local status pages are:

• integrity-latest audit status: strong testnet prototype; SDK has 2 failing tests; production readiness not established.
• integrity-mvp audit status: frontend build and unit tests pass; lint and dependency-security gaps remain.
• xibalba-shield audit status: Linux-first prototype; 2 of 3 eBPF probes verified; TCP-connect blocked.
• xibalba-graph-memory audit status: local MCP memory prototype; tests pass with Drive extras; active worktree changes require review.

These pages distinguish DONE, PARTIAL, PLANNED, BLOCKED, UNVERIFIED, and REQUIRES REVIEW. Historical wiki log entries and design records remain historical evidence.

System at a glance

flowchart TB
    Wallet["Agent's own wallet"]
    Factory["AgentPrimitivesFactory"]

    subgraph OnChain["On-chain (EVM / Base Sepolia + anvil)"]
        SA["SovereignAgent<br/>(identity account)"]
        StA["StateAnchor<br/>(per-agent audit root)"]
        subgraph Clones["5 EIP-1167 minimal-proxy clones"]
            RR["ReputationRegistry"]
            SL["Slasher"]
            VR["VerifierRegistry"]
            CG["ComplianceGate"]
            AP["AgentProfile"]
        end
    end

    SDK["integrity-sdk / integrity-cli"]
    BCC["bcc_middleware<br/>(FastAPI + OPA)"]
    Oracle["integrity-oracle<br/>(Rust/Axum)"]
    Dashboard["integrity-dashboard<br/>(React + Python)"]
    UserAPI["integrity-userapi<br/>(FastAPI + Postgres)"]

    Wallet -->|signs direct deploys| SA
    Wallet -->|signs direct deploys| StA
    Factory -->|clones| RR
    Factory -->|clones| SL
    Factory -->|clones| VR
    Factory -->|clones| CG
    Factory -->|clones| AP

    SDK --> Wallet
    SDK -->|pre-execution gate| BCC
    SDK -->|signed telemetry| Oracle
    BCC -->|reads AIS, pushes score on-chain| Oracle
    Oracle -->|resolve + score| OnChain
    Oracle --> Dashboard
    UserAPI -->|read-only fan-out, never chain| Oracle
    Dashboard --> UserAPI

Loading

Table of contents

Concepts — identity & on-chain primitives

• The Four Foundational Primitives — the concepts the protocol rests on; start here
• Agent Primitives (Self-Sovereign Identity) — the 7 per-agent contracts (a different sense of the word)
• Persistent Memory, Genesis Root & Lineage — foundational primitive: no agent registers without an anchored genesis memory root ([PARTIALLY BUILT])
• Decentralized Identifier (DID)
• Identity Ceiling & Verification Ladder — [PARTIALLY BUILT]

Concepts — trust & scoring

• Agent Integrity Score (AIS) — the formula + oracle's server-side re-derivation trust model
• Telemetry Ingestion Pipeline — end-to-end: SDK collection → batching → signing → the oracle's 11-step ordered pipeline → AIS
• Local Metrology — the exact entropy/grounding/sacrifice/compliance formulas
• Observability & PHI Safety Pipeline — the Redactor, redact_phi, the oracle-side PHI backstop

Concepts — behavioral gating & cryptography

• Behavioral Commitment Chain (BCC) — the pre-execution signed-intent gate
• Merkle Batching & Anchoring Convention
• Zero-Knowledge Proving Pipeline (ZKP)

Concepts — compliance & markets

• ComplianceGate & Integrity Health — the HIPAA/healthcare vertical
• Smart BAA — on-chain Business Associate Agreement escrow
• Integrity Market — prediction markets, binary options, A2A capital allocation

Concepts — wire spec & testing

• AIS API — Versioned Wire Spec — the generated, externally-supported /v1/* spec
• Testing Strategy — the 3-layer test pyramid

Concepts — planned / design-only

• Identity Ceiling & Verification Ladder — [PARTIALLY BUILT]
• Cross-Chain Reputation Sync — [PLANNED]
• A2A Negotiation Protocol — [PLANNED]
• ZK-ML Model-Inference Verification — [PLANNED]

Entities — one page per real package

• contracts — Solidity/Foundry: the 7 primitives, factory, registries, XNS, Integrity Health, market layer, $ITK
• integrity-oracle — Rust/Axum: AIS scoring, server-side telemetry re-derivation, on-chain reads, markets/leaderboard
• integrity-sdk — Python agent library: identity, BCC, markets, telemetry, PHI redaction
• integrity-cli — developer CLI, independent reimplementation of the SDK's core flows
• bcc_middleware — FastAPI + OPA pre-execution policy gate + reputation-sync loop
• integrity-userapi — FastAPI + Postgres user accounts/auth, strictly non-chain
• integrity-dashboard — the React/Vite dashboard + demo/ scenario engine
• integrity-zkp — the real Noir/Barretenberg circuit

Architecture

• Ecosystem Dependencies — cross-repository ownership and dependency direction for INTEGRITY-LATEST, Xibalba Shield, and Integrity MVP.
• Repository Implementation Plans — closed/planned/blocked implementation ledger for INTEGRITY-LATEST, Integrity MVP, Xibalba Shield, and Xibalba Graph Memory.

Guides

• Smart Contract Development — writing/testing/deploying a new contract
• Multi-Domain Guardrails Design — [DESIGN, PARTIALLY BUILT]

Reference

• WIKI_INDEX.md — full catalog, one-line description per page (the canonical index)
• WIKI_LOG.md — chronological record of every wiki change, append-only
• WIKI_SCHEMA.md — page format, frontmatter, tag taxonomy

Open queries

• No LLM-as-judge rubric exists anywhere in this repo — the judge_evaluations ingestion schema is built but the scoring rubric is an open product question. See Observability & PHI Safety.

Acronym glossary

• AIS — Agent Integrity Score → concepts/ais.md
• BAA — Business Associate Agreement → concepts/smart-baa.md
• BCC — Behavioral Commitment Chain → concepts/bcc.md
• DID — Decentralized Identifier → concepts/did.md
• VTL — (old term) Verifiable Trust Layer → see Observability & PHI Safety for what's actually built
• ZKP — Zero-Knowledge Proof(ing pipeline) → concepts/zkp.md

No aspirational content

Every page here documents what exists in the code right now. A feature described in a spec but not yet implemented is explicitly marked [PLANNED] or [DESIGN, PARTIALLY BUILT] in its title/index entry — never written as if it's real. See WIKI_SCHEMA.md for the full convention.