ethereum-wingman
π‘ Summary
Ethereum Wingman is an AI-powered tutor for developing Ethereum dApps and smart contracts.
π― Target Audience
π€ AI Roast: βPowerful, but the setup might scare off the impatient.β
Risk: Medium. Review: shell/CLI command execution; outbound network access (SSRF, data egress); API keys/tokens handling and storage; filesystem read/write scope and path traversal. Run with least privilege and audit before enabling in production.
Ethereum Wingman
A comprehensive Ethereum development tutor and guide built as an Agent Skill. Teaches smart contract development through SpeedRun Ethereum challenges, Scaffold-ETH tooling, and security best practices.
Quick Start
Build an Ethereum dApp from scratch with AI assistance:
# 1. Create a new project folder mkdir my-defi-project cd my-defi-project # 2. Install the Ethereum Wingman skill npx skills add austintgriffith/ethereum-wingman # 3. Open in Cursor (or your AI-enabled editor) cursor .
Then just tell the AI what you want to build:
"Help me build a dApp where users can stake ETH and earn rewards"
"Create a token with buy/sell functionality like SpeedRun Ethereum Challenge 2"
"Build an NFT collection with on-chain SVG art"
The Ethereum Wingman will:
- ποΈ Scaffold a Scaffold-ETH 2 project for you
- β οΈ Warn you about critical gotchas (token decimals, reentrancy, etc.)
- π Guide you on security best practices
- π Reference SpeedRun Ethereum challenges for learning
What is Ethereum Wingman?
Ethereum Wingman is a knowledge base and prompt system that helps AI agents assist developers learning Ethereum development. It covers:
- SpeedRun Ethereum Challenges: TLDR modules for all 12 challenges
- Scaffold-ETH 2 Integration: Tooling docs, hooks reference, fork workflows
- DeFi Protocols: Uniswap, Aave, Compound patterns
- ERC Standards: Comprehensive guides for ERC-20, 721, 1155, 4626
- Security: Critical gotchas, historical hacks, pre-production checklist
Installation
Via skills.sh (Recommended)
npx skills add austintgriffith/ethereum-wingman
This works with Cursor, Claude Code, Codex, OpenCode, and other AI coding agents.
Manual Installation
For Cursor:
Copy .cursorrules to your project root or add to your global Cursor rules.
For Claude Code:
Reference the CLAUDE.md file in your project instructions.
As MCP Integration: The skill.json manifest describes capabilities that can be integrated with MCP-compatible agents.
Directory Structure
ethereum-wingman/
βββ skills/
β βββ ethereum-wingman/ # skills.sh compatible package
β βββ SKILL.md # Skill definition with frontmatter
β βββ AGENTS.md # Full compiled instructions
β βββ metadata.json # Skill metadata
β βββ README.md # Skill documentation
β βββ scripts/ # Helper scripts
β β βββ init-project.sh
β β βββ check-gotchas.sh
β βββ references/ # Key knowledge files
βββ knowledge/
β βββ challenges/ # 12 SpeedRun Ethereum challenge modules
β βββ protocols/ # DeFi protocol documentation
β βββ standards/ # ERC standards
β βββ foundations/ # Core concepts
β βββ gotchas/ # Security knowledge
βββ tools/
β βββ scaffold-eth/ # Scaffold-ETH 2 documentation
β βββ deployment/ # Deployment patterns
β βββ security/ # Security tools
βββ prompts/ # AI agent prompts
βββ AGENTS.md # Symlink to skills/ethereum-wingman/AGENTS.md
βββ skill.json # Legacy skill manifest
βββ .cursorrules # Cursor IDE integration
βββ CLAUDE.md # Claude Code integration
Key Concepts Covered
π¨ The Most Important Concept
NOTHING IS AUTOMATIC ON ETHEREUM.
Smart contracts cannot execute themselves. For any function that "needs to happen":
- Make it callable by ANYONE (not just admin)
- Give callers a REASON (profit, reward, their own interest)
- Make the incentive SUFFICIENT to cover gas + profit
The Wingman will always ask: "Who calls this function? Why would they pay gas?"
Critical Gotchas
Every Ethereum developer must know:
- Token Decimals Vary: USDC = 6, WBTC = 8, most = 18
- Approve Pattern Required: Contracts need approval before transferFrom
- Reentrancy Attacks: Always use Checks-Effects-Interactions + ReentrancyGuard
- Oracle Manipulation: Never use DEX spot prices
- No Floating Point: Use basis points (500/10000 = 5%)
- Nothing is Automatic: Design incentives for function callers
- Vault Inflation Attack: Protect first depositors
SpeedRun Ethereum Challenges
Each challenge teaches a key blockchain concept:
| Challenge | Concept | |-----------|---------| | Simple NFT | ERC-721, minting, metadata | | Decentralized Staking | Coordination, deadlines, escrow | | Token Vendor | ERC-20 approve pattern | | Dice Game | Randomness vulnerabilities | | DEX | AMM, constant product formula | | Oracles | Price feeds, manipulation resistance | | Lending | Collateralization, liquidation | | Stablecoins | Pegging mechanisms | | Prediction Markets | Outcome resolution | | ZK Voting | Privacy-preserving governance | | Multisig | Threshold signatures | | SVG NFT | On-chain generative art |
Usage Examples
Teaching Mode
Ask questions like:
- "How does the ERC-20 approve pattern work?"
- "Explain the constant product formula in AMMs"
- "What are the security considerations for a lending protocol?"
Code Review Mode
Submit code for review:
- "Review this withdrawal function for vulnerabilities"
- "Check this vault contract for inflation attacks"
- "Audit this oracle integration"
Build Mode
Get help building:
- "Help me build a token with buy/sell functionality"
- "Set up a staking contract with rewards"
- "Create an NFT minting page"
Debug Mode
Troubleshoot issues:
- "Why is my transaction reverting?"
- "How do I fix 'insufficient allowance' errors?"
- "Debug this reentrancy issue"
Contributing
To add new content:
- Add markdown files to appropriate directory
- Follow existing format (TLDR, code examples, security notes)
- Update skill.json if adding new capabilities
- Test with AI agent to ensure clarity
License
MIT License - Use freely for learning and building.
Credits
Built for the BuidlGuidl community.
Integrates knowledge from:
Pros
- Comprehensive learning resources
- AI assistance for coding
- Focus on security best practices
- Integration with popular tools
Cons
- Requires familiarity with Ethereum
- Limited to specific AI tools
- May not cover all edge cases
- Dependency on external resources
Related Skills
pytorch
SβIt's the Swiss Army knife of deep learning, but good luck figuring out which of the 47 installation methods is the one that won't break your system.β
agno
SβIt promises to be the Kubernetes for agents, but let's see if developers have the patience to learn yet another orchestration layer.β
nuxt-skills
SβIt's essentially a well-organized cheat sheet that turns your AI assistant into a Nuxt framework parrot.β
Disclaimer: This content is sourced from GitHub open source projects for display and rating purposes only.
Copyright belongs to the original author austintgriffith.
