Does this skill require specific environment settings?

Yes, you must set CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 in your .claude/settings.json or shell environment to enable the team orchestration features.

What happens if I already have an existing audit directory?

The skill includes a pre-flight discovery phase that detects previous runs and offers to archive, overwrite, or cancel to ensure your audit data remains organized.

How does the human-in-the-loop process work?

The skill includes interactive stages where the lead agent asks for design decisions (Stage 0) and presents findings for you to classify as valid, disputed, or ignored (Stage 4).

Can it detect reentrancy and math errors?

Yes, specific stages of the audit are dedicated to reentrancy, trust boundaries, math, rounding, and state consistency, often supplemented by Slither static analysis.

What makes this different from a standard Solidity audit skill?

This skill uses an 'agent team' approach, meaning multiple specialized AI agents work in parallel with inter-agent messaging and shared task lists, providing a more thorough and multi-perspectival review than a single agent.

Solidity Function Audit Team

Name: Solidity Function Audit Team
Author: gonzaloetjo

bygonzaloetjo

0•

セキュリティとテスト

Orchestrates a multi-agent team to perform deep, structured security audits of Solidity smart contracts with human-in-the-loop oversight.

The Solidity Function Audit Team skill transforms Claude Code into a comprehensive security auditing powerhouse. By leveraging experimental agent team capabilities, it distributes the auditing process across specialized agents who handle state variable mapping, access control analysis, and adversarial sequence modeling. The workflow is uniquely designed to be context-aware, starting with an interactive session to capture developer design intent and concluding with a human-guided review stage to classify and refine findings. It integrates seamlessly with Slither for static analysis, providing a rigorous, multi-layered approach to identifying vulnerabilities in complex DeFi protocols and smart contract systems.

主な機能

01Multi-agent collaboration using inter-agent messaging and shared task dependencies.

020 GitHub stars

03Automated Slither static analysis integration for initial vulnerability surfacing.

04Interactive Stage 0 design decision capture to align AI analysis with developer intent.

05Structured human-in-the-loop review (Stage 4) and dispute re-evaluation (Stage 5).

06Domain-based function grouping for specialized, deep-dive architectural reviews.

ユースケース

01Performing deep-dive logic audits on specific functional domains within a protocol.

02Automating multi-perspective security reviews during the CI/CD development lifecycle.

03Conducting pre-deployment security assessments for complex DeFi smart contracts.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add gonzaloetjo/solidity-audit-skills solidity-function-audit-team

For use in Claude.ai and ChatGPT

Download Skill