deep-dive

Audit codebase scopes for security flaws, logic errors, and performance bottlenecks.

437|45|Updated Jan 27, 2026
One-click install
npx skills add https://github.com/ZaxbyHub/opencode-swarm --skill deep-dive-zaxbyhub
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deep-dive
Source: https://github.com/ZaxbyHub/opencode-swarm/tree/main/.opencode/skills/deep-dive
Command: npx skills add https://github.com/ZaxbyHub/opencode-swarm --skill deep-dive-zaxbyhub

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill eliminates the risk of missed critical code issues, unverified audit findings, and accidental code mutations during manual or single-pass codebase reviews, ensuring only thoroughly vetted findings are surfaced.

Core Features & Use Cases

  • Parallel Explorer Waves: Dispatches multiple specialized audit lanes (security, performance, UX, architecture etc.) to cover different aspects of the target scope simultaneously, with no overlapping file assignments.
  • Dual Independent Review: All candidate findings from explorer waves are verified by two separate reviewer agents to eliminate false positives and ensure accuracy.
  • Critic Challenge for High-Severity Findings: High and critical verified findings undergo sequential critic review to confirm root cause accuracy and appropriate severity rating.
  • Use Case: An engineering team preparing to release a new authentication module can run a deep-dive audit to catch security vulnerabilities, logic gaps, and performance issues before deployment, with zero risk of modifying the existing codebase.

Quick Start

Use the deep-dive skill to run a full read-only audit of the user authentication module in the current repository, using the full audit profile and outputting results as markdown.

Frequently Asked Questions about deep-dive

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

FAQPage Schema
How do I perform a security audit on my codebase without modifying any files?

A read-only security audit analyzes your codebase scopes to identify flaws without mutating existing code. It dispatches parallel explorer waves to cover security and performance simultaneously, ensuring thorough pre-deployment review with zero risk of accidental changes.

What's the best way to verify code review findings and eliminate false positives?

To eliminate false positives during a code review, use dual independent reviewer verification where two separate agents check all candidate findings. High-severity issues then undergo sequential critic challenge to confirm root cause accuracy and appropriate severity rating.

How does parallel code review work for technical debt assessment?

Parallel code review dispatches multiple specialized audit lanes covering architecture, performance, and security simultaneously with no overlapping file assignments. This approach catches logic errors and design gaps across different aspects of the target scope at once for comprehensive technical debt assessment.

Can I run a pre-deployment code review to catch both logic errors and performance bottlenecks?

Yes, pre-deployment code review applies high-rigor audits to specified scopes to identify logic errors, performance bottlenecks, and security vulnerabilities. It uses parallel explorer waves and dual verification to ensure only thoroughly vetted findings are surfaced before release.

Does a read-only code audit require any specific dependencies or components?

No dependencies or components are required to run a read-only code audit. The process applies parallel explorer wave execution, dual independent reviewer verification, and sequential critic challenge internally to ensure audit accuracy for any software project scope.

When should I not use a single-pass codebase review instead of a deep-dive audit?

Avoid single-pass codebase review when preparing to release critical modules like authentication, where missing security vulnerabilities or logic gaps is costly. A deep-dive audit eliminates the risk of unverified findings and missed critical issues through multi-stage verification.