How does this skill use multiple AI models?

It orchestrates Claude subagents for internal codebase analysis, Gemini CLI for real-time web research, and Codex CLI for GitHub pattern matching and deep reasoning.

Is the analysis data saved for later use?

Yes, all research logs, plans, and reports are saved to an external .analysis/ directory, making the process fully resumable and audit-ready.

Do I need Gemini or Codex installed to use this skill?

The skill works best with Gemini and Codex CLIs installed, but it features a Claude-only fallback mode if those tools are unavailable.

How is the quality score calculated?

A specialized Verification Agent grades the output on a 100-point rubric covering comprehensiveness, accuracy, specificity, actionability, and consistency.

What is the Progressive Disclosure strategy?

It is a file-reading method that starts with metadata (glob) and patterns (grep) before reading full files (read) to minimize token usage and speed up analysis.

Multi-AI Research & Analysis

Name: Multi-AI Research & Analysis
Author: adaptationio

byadaptationio

安全与测试

Conducts deep, multi-perspective technical research and analysis using integrated Claude, Gemini, and Codex agents for verified, production-ready results.

关于

Multi-AI Research is a high-rigor analysis skill that orchestrates Claude subagents alongside Gemini and Codex CLIs to provide exhaustive technical evaluations. It moves beyond simple querying by implementing a 5-phase pipeline—planning, parallel research, deep analysis, synthesis, and iterative verification—to ensure 100% citation coverage and architectural depth. Whether auditing security, mapping complex system architectures, or identifying performance bottlenecks, this skill uses an external memory architecture to maintain state across sessions and cross-validates findings between three distinct AI perspectives to eliminate hallucinations and fill information gaps.

主要功能

Quality-Gated Iteration: Automatically refines analysis until it achieves a verified quality score of 95/100 or higher.
External Memory Architecture: Saves state to local files, allowing for multi-session work and context-window resilience.
Progressive Disclosure Strategy: Efficiently analyzes large codebases using a 3-level approach (Glob > Grep > Read).
Multi-Perspective Verification: Cross-validates findings between Claude, Gemini (web), and Codex (GitHub patterns).
100% Citation Coverage: Ensures every claim is backed by specific file line numbers or external source URLs.
0 GitHub stars

使用场景

Deep architecture reviews and component mapping for legacy or complex microservices systems.
Comprehensive security audits and vulnerability assessments with OWASP Top 10 coverage.
Performance bottleneck identification and algorithm complexity analysis with actionable optimization steps.

关于

主要功能

Quality-Gated Iteration: Automatically refines analysis until it achieves a verified quality score of 95/100 or higher.
External Memory Architecture: Saves state to local files, allowing for multi-session work and context-window resilience.
Progressive Disclosure Strategy: Efficiently analyzes large codebases using a 3-level approach (Glob > Grep > Read).
Multi-Perspective Verification: Cross-validates findings between Claude, Gemini (web), and Codex (GitHub patterns).
100% Citation Coverage: Ensures every claim is backed by specific file line numbers or external source URLs.
0 GitHub stars

使用场景

Deep architecture reviews and component mapping for legacy or complex microservices systems.
Comprehensive security audits and vulnerability assessments with OWASP Top 10 coverage.
Performance bottleneck identification and algorithm complexity analysis with actionable optimization steps.