What is the typical duration for a full analysis?

Depending on project size, a comprehensive multi-AI analysis typically takes between 45 and 90 minutes to ensure rigorous verification and iterative refinement.

How does the '100% citation coverage' work?

The system mandates that every claim in the final synthesis report is backed by a specific file:line reference or a verified URL captured during the parallel research phase.

How does this prevent excessive token usage?

It uses a Progressive Disclosure strategy, analyzing file structures and grep patterns first, and only reading the full content of critical files identified during discovery.

Can I resume a research session if it's interrupted?

Yes. All state is saved in a local '.analysis/' directory, allowing the skill to survive context window limits and resume from any checkpoint.

Does this skill require Gemini and Codex CLI to be installed?

While it performs best with Gemini and Codex CLIs for multi-perspective analysis, the skill includes a Claude-only fallback mode if the other tools are unavailable.

Multi-AI Research & Analysis

Name: Multi-AI Research & Analysis
Author: adaptationio

byadaptationio

0•

Security & Testing

Conducts deep, multi-perspective technical audits and research by orchestrating Claude, Gemini, and Codex AI systems.

This skill transforms Claude Code into a comprehensive research powerhouse by coordinating three distinct AI systems—Claude, Gemini, and Codex—to perform rigorous technical analysis. It specializes in cross-verifying findings to ensure 100% citation coverage and high-quality outputs for security audits, architectural reviews, and code quality assessments. By using an external memory architecture and a standardized 5-phase pipeline, it maintains state across sessions and provides actionable, evidence-based recommendations that far exceed the depth and accuracy of a single-model analysis.

Key Features

01Automated Quality Scoring: Rigorous 3-pass verification system ensuring ≥95/100 quality and 100% citation accuracy.

02Progressive Disclosure: Efficient codebase scanning using metadata and grep patterns to minimize token usage.

03Multi-AI Orchestration: Leverages Claude, Gemini CLI, and Codex CLI for diverse research perspectives.

045-Phase Pipeline: Systematic workflow covering planning, parallel research, deep analysis, synthesis, and verification.

050 GitHub stars

06External Memory Architecture: Saves state to local files for resumable, multi-session analysis without context loss.

Use Cases

01Performing comprehensive security audits and vulnerability assessments against OWASP standards.

02Benchmarking code quality and identifying refactoring opportunities with multi-source verification.

03Conducting deep architectural reviews and system design mapping for complex or legacy codebases.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add adaptationio/skrillz multi-ai-research

For use in Claude.ai and ChatGPT

Download Skill