What is Mnemom?

Mnemom is a trust plane for the agentic internet, designed to ensure AI agents operate within declared boundaries and are cryptographically verifiable. It provides a stack of protocols including the Agent Alignment Protocol (AAP), Agent Integrity Protocol (AIP), and Card Lifecycle & Policy Intelligence (CLPI), along with a cross-tenant defensive network called Mnemom AEGIS. The platform enables organizations to ship agents into regulated workflows with audit-ready evidence and trust ratings.

How to use Mnemom?

To use Mnemom, start by publishing an Alignment Card under the Agent Alignment Protocol (AAP) that declares your agent's intent, permitted tools, data boundaries, and escalation contract. Then, integrate the Safe House to screen every transaction at four checkpoints (front door, back door, inside.autonomy, inside.integrity). Verify cryptographic proofs using AIP and CLPI, and monitor Trust Ratings that refresh every 6 hours. You can also install the SDK and start verifying in under 5 minutes, or explore the Showcase for live simulations.

Core features of Mnemom

Alignment Cards: Declare agent intent, autonomy bounds, and audit commitments under AAP
Agent Integrity Protocol (AIP): Checkpoints every thinking block before action execution, providing verdicts like clear, review needed, or boundary violation
Card Lifecycle & Policy Intelligence (CLPI): Five-phase governance with on-chain anchoring on Base L2 and OpenTelemetry observability
Mnemom AEGIS: Cross-tenant defensive network that screens transactions at four checkpoints and propagates signed Managed Rules
Trust Ratings: Cryptographic reputation scores from 0-1000 based on observed behavior, with multi-agent coherence
Supply-chain detection: Substrate fingerprinting to catch anomalies across customers
Open protocols: AAP and AIP are open source under Apache 2.0