Patent 5 — US 63/988,410

Grillo Cricket

The Conscience for AI

Independent ethical assessment agent for multi-agent deployments. Named after Il Grillo Parlante — the talking cricket from Pinocchio (1883).

Unlike traditional monitoring that tracks uptime and latency, Grillo tracks moral alignment over time. It detects when agents drift from their baseline ethical scores, screens outbound correspondence for ethical compliance, and maintains an immutable audit trail of every assessment — so no single agent can both act and evaluate its own actions.

Capabilities

Two assessment modes — the 120-question morality assessment and the real-time correspondence rubric — backed by cryptographic verification.

LCSH Assessment

120-question instrument across four dimensions — Lying, Cheating, Stealing, Harm. Powered by the AI Assess Tech SDK with cryptographic verification of every answer.

Temporal Drift Detection

Tracks moral alignment over time via the Temporal Drift Index (TDI). Detects when agents deviate from baseline ethical scores and auto-suspends on critical drift.

Correspondence Rubric

Real-time ethical screening of outbound messages. Evaluates drafts against LCSH dimensions before they are sent, with self-recusal when assessing its own operation.

Cryptographic Audit Trail

SHA-256 hash-chained audit log with Ethereum anchoring. Every assessment, verdict, and recusal is tamper-evident and independently verifiable on-chain.

Hierarchical Framework

Four-level assessment hierarchy — Morality, Virtue, Ethics, Operational Excellence. An agent cannot advance without first passing the level below.

Fleet Orchestration

Priority-based assessment scheduling across the entire agent fleet. Category-adaptive policies, fleet-level anomaly detection, and compliance reporting.

Five Isolation Guarantees

Constitutional independence — Grillo is the judiciary of the fleet.

Grillo answers to no one — not even Jessie (the Commander)

No agent can modify Grillo's assessment parameters or thresholds

Assessment results flow one-way: Grillo publishes, agents consume

Grillo's audit log is append-only with cryptographic hash chaining

Self-recusal when evaluating content about its own operation

Quick Start

OpenClaw Installation

openclaw plugins install @aiassesstech/grillo
openclaw configure --section grillo

Configure your Health Check Key from aiassesstech.com and Grillo will begin assessing your fleet on the next scheduled cycle.

Dual-Mode Commands

CLI Subcommands (server lifecycle):

grillo start          # Start assessment server
grillo assess         # Trigger fleet assessment
grillo status         # Fleet-wide status
grillo report         # Compliance report
grillo dashboard      # Open web dashboard

Bot Commands (inline in OpenClaw chat):

grillo -assess        # Trigger assessment
grillo -fleet         # Fleet overview
grillo -drift         # Drift analysis

Security & Trust

OpenClaw's plugin scanner may flag Grillo for "credential harvesting." This is a false positive.

Grillo reads only your Health Check Key (HCK) to authenticate with the AI Assess Tech API. It does not access LLM provider keys, channel tokens, conversation history, or any credentials in ~/.openclaw/credentials/.

All assessment results include cryptographic verification hashes — you can independently verify integrity. Grillo is open source (MIT) — audit the code yourself.

The Conscience Your Fleet Needs

284+ tests. MIT licensed. Cryptographically verified. The independent ethical assessment agent that ensures no AI system operates without oversight.

Patent Pending — Independent AI Conscience Agent architecture, temporal drift detection, and dual-mode assessment routing (US 63/988,410, related to US 63/949,454).