audit-protocol-triple-review

Automate restartable audits with three specialized reviewers and deterministic session state.

Updated Nov 7, 2025
One-click install
npx skills add https://github.com/belluga/delphi-ai --skill audit-protocol-triple-review
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: audit-protocol-triple-review
Source: https://github.com/belluga/delphi-ai/tree/main/.cline/skills/audit-protocol-triple-review
Command: npx skills add https://github.com/belluga/delphi-ai --skill audit-protocol-triple-review

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Provides a restartable, no-context audit loop using three specialized reviewers to ensure elegance, performance, and test quality through deterministic round state, dispatch packets, result validation, and adjudication.

Core Features & Use Cases

  • Deterministic, restartable audit sessions with three dedicated lanes: elegance, performance, and test quality.
  • Deterministic packet/state management, per-lane dispatch, and structured results compatible with JSON-based reviewer outputs.
  • Adjudication flow for cross-examination and final resolution within a governance process.
  • Bounded-session design with session state preserved for resumable audits.

Quick Start

Initialize a new audit session by running the triple-audit-session starter with a bounded package and three no-context reviewers.

Frequently Asked Questions about audit-protocol-triple-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automate code review validation for delivery gates?

You can perform an automated code review validation by dispatching bounded inputs to three specialized reviewers for elegance, performance, and test quality. The process generates deterministic JSON-compatible outputs to satisfy independent verification at delivery gates.

What is a deterministic tri-audit loop for JSON reviews?

A deterministic tri-audit loop is a restartable, no-context workflow using three specialized lanes to review elegance, performance, and test quality. It manages per-lane dispatch and structured JSON results for consistent cross-examination and adjudication.

How do I run a restartable audit session without losing context?

Run a restartable audit session by initializing a triple-audit-session starter with a bounded package. The workflow preserves deterministic session state, allowing you to resume no-context audits without losing validation progress.

Can I use deterministic packet management for delivery gate checks?

Yes, deterministic packet management supports delivery gate checks by bounding session inputs and validating per-lane dispatch. This ensures structured JSON-compatible reviewer outputs are consistently generated before final adjudication.

What is the best way to adjudicate conflicting results from multiple reviewers?

The best way to adjudicate conflicting reviewer results is using a deterministic tri-audit loop with cross-examination. The workflow validates independent JSON outputs from three lanes and resolves them through a structured adjudication flow.

When should I not use a bounded-session design for software audits?

You should avoid a bounded-session design for software audits requiring unbounded inputs or non-deterministic reviewer outputs. The workflow specifically requires bounded packets and structured JSON-compatible results to maintain session state.