Web3 Godmode Config

Claude Code configuration for web3 polyglot engineers. Solidity + Rust + TypeScript/React, multi-chain.

Designed to work standalone or alongside the compound-engineering plugin for maximum capability.

Prerequisites

• Claude Code CLI installed
• Python 3 (for hooks)
• jq (for todo-enforcer hook): brew install jq
• Optional: compound-engineering plugin for 27 additional review/analysis agents

Installation

Quick Install

git clone https://github.com/0xinit/web3-godmode-config.git
cd web3-godmode-config
./install.sh

The installer will:

1. Check prerequisites
2. Detect compound-engineering plugin
3. Backup existing ~/.claude/ config
4. Copy all skills to ~/.claude/godmode-store/ and symlink enabled ones to ~/.claude/skills/
5. Install the godmode CLI to ~/.claude/bin/
6. Register hooks in settings.json
7. Verify installation

After install, add to your shell profile:

export PATH="$HOME/.claude/bin:$PATH"

Manual Install

# Copy everything to ~/.claude/
cp CLAUDE.md ~/.claude/
cp -r agents/ ~/.claude/agents/
cp -r skills/ ~/.claude/godmode-store/   # store, not skills/
cp -r rules/ ~/.claude/rules/
cp -r commands/ ~/.claude/commands/
cp -r hooks/ ~/.claude/hooks/
cp godmode ~/.claude/bin/godmode && chmod +x ~/.claude/bin/godmode
cp godmode-skills.conf ~/.claude/godmode-skills.conf

# Make hooks executable
chmod +x ~/.claude/hooks/*.py ~/.claude/hooks/*.sh

# Symlink just the skills you want
ln -s ~/.claude/godmode-store/rigorous-coding ~/.claude/skills/rigorous-coding

Then manually add hooks to ~/.claude/settings.json (see install.sh for hook config).

Skill Management (godmode CLI)

Skills are stored in ~/.claude/godmode-store/ and symlinked on-demand into ~/.claude/skills/. This prevents overflowing Claude Code's ~16K skill description budget when combined with plugins.

godmode list                  # show all 18 skills, enabled/disabled
godmode enable megaeth        # fuzzy match → enables web3-megaeth
godmode enable solidity       # matches solidity-security + web3-solidity-patterns
godmode disable megaeth       # remove from active skills
godmode enable-all            # symlink everything
godmode disable-all           # clear all active skills
godmode status                # show budget usage (chars used / 16K limit)
godmode profile solidity      # enable a predefined skill group

Profiles (defined in godmode-skills.conf):

Profile	Skills
solidity	rigorous-coding, solidity-security, web3-solidity-patterns, web3-foundry, web3-hardhat, web3-eip-reference, smart-contract-audit
react	rigorous-coding, react-useeffect, vercel-react-best-practices, web3-frontend, web3-privy
rust	rigorous-coding, rust-patterns, web3-solana-simd
fullstack	rigorous-coding, planning-with-files, react + solidity essentials

By default, only planning-with-files and rigorous-coding are enabled on install. Edit godmode-skills.conf to change defaults, or use the CLI at runtime.

Customization

Adding a New Chain

mkdir -p ~/.claude/skills/web3-YOUR-CHAIN/

Create a SKILL.md inside:

---
name: web3-your-chain
description: Your Chain development reference. Use when deploying to or building on Your Chain.
---

# Your Chain Reference

## Chain Configuration
| Property | Value |
|----------|-------|
| Chain ID | ... |
| RPC URL | ... |
| Explorer | ... |
| Faucet | ... |

## Deployment Patterns
...

Then add it to two places:

1. CLAUDE.md — add a row to the skill trigger table
2. rules/forge.md — add a row to the chain configuration table

Modifying Rules

Rules in rules/ are auto-loaded based on the file path you're editing. The paths: frontmatter controls when they activate:

---
paths: "**/*.sol"  # Activates when editing any .sol file
---

Examples:

# Only activate in a specific directory
paths: "contracts/**/*.sol"

# Activate for multiple extensions
paths: "**/*.{ts,tsx}"

# Activate for test files only
paths: "**/*.{test,spec}.ts, **/__tests__/**"

# Activate for config files
paths: "foundry.toml, hardhat.config.ts"

To create a new rule, add a .md file to rules/ with the frontmatter and your rules below it.

Adding a New Keyword Mode

Edit hooks/keyword-detector.py and add an entry to the MODES dict:

"your-mode": {
    "pattern": r"\b(your-keyword|alias)\b",
    "context": """[YOUR MODE ACTIVATED]

Instructions that get injected into Claude's context when triggered...
"""
},

Adjusting the Comment Checker

Edit hooks/check-comments.py:

• MAX_COMMENT_RATIO — change the threshold (default 25%)
• VALID_PATTERNS — add patterns that should NOT be flagged (e.g., your custom annotations)
• CODE_EXTENSIONS — add/remove file extensions to check

Adding a New Command

Create a .md file in commands/:

---
allowed-tools: Read, Glob, Grep, Bash
argument-hint: [your-args]
description: What this command does
---

Instructions for Claude when the user runs /your-command...

Usage: /your-command [args] in any Claude Code session.

Adding a New Skill

Create a directory in skills/ with a SKILL.md:

---
name: your-skill
description: When to use this skill. Be specific — Claude uses this to decide when to auto-invoke.
---

# Your Skill

Reference content, patterns, code examples...

Set disable-model-invocation: true in the frontmatter if the skill should only run when manually invoked (not auto-triggered).

Usage Examples

Keyword Modes

Type these words anywhere in your prompt to activate specialized behavior:

ultrawork implement the staking contract

Activates maximum parallel execution, comprehensive planning, todo tracking.

analyze why the approve function reverts on Base

Activates deep investigation protocol with multi-phase analysis.

security review the vault contract

Routes to security scanning with Solidity-specific checks (reentrancy, CEI, access control).

deploy the NFT contract to Base mainnet

Activates deployment protocol — confirms chain, runs pre-deploy checks, verifies after.

All keyword modes:

Keyword	What Happens
ultrawork / ulw	Maximum parallel execution, comprehensive planning
search / find	Exhaustive multi-angle search
analyze / debug	Deep investigation with evidence gathering
think deeply	Extended reasoning with trade-off analysis
refactor	Incremental changes, behavior preservation, tests first
review	Routes to appropriate reviewer agent
test	BDD structure, edge cases, fuzz testing
optimize	Gas optimization (Solidity) or bundle/perf (React)
security / audit	Security scan (Solidity-specific or general)
deploy	Deployment protocol with chain awareness

Commands

/audit contracts/Vault.sol

Runs a structured security audit: static analysis, access control, reentrancy, economic vectors, gas DoS, events, NatSpec.

/deploy-check base MyToken

Pre-deployment verification: chain config, constructor args, proxy setup, post-deploy checks. Returns GO/NO-GO.

/interview docs/feature-spec.md

Interactive interview to flesh out a plan or spec. Asks deep questions about implementation, UX, tradeoffs.

Skills Auto-Invoke

Skills trigger automatically based on what you're doing:

• Edit a .sol file → Solidity rules load + security skill available
• Write React hooks → useEffect patterns and Vercel best practices activate
• Ask about an EIP → EIP reference skill provides inline summaries
• Work on Anchor program → Rust patterns skill with account validation checks

Compound-Engineering Integration

If you have the compound-engineering plugin installed, this config routes to its agents automatically:

Need	Godmode Provides	Compound Provides
Solidity security	/solidity-security skill	security-sentinel agent
Code review	Rules + patterns	kieran-*-reviewer agents
Performance	Gas optimization skill	performance-oracle agent
Architecture	—	architecture-strategist agent
PR workflow	—	/workflows:review command
Frontend	Web3 frontend patterns	frontend-design skill
Git commits	—	gitbahn commands

Both are triggered automatically via the CLAUDE.md delegation table — no manual routing needed.

What's Included

Skills (18)

Skill	What It Does
rigorous-coding	Pre/post implementation checks, error handling, naming
planning-with-files	Manus-style persistent planning with markdown files
react-useeffect	When NOT to use useEffect + better alternatives
vercel-react-best-practices	45 React/Next.js performance optimization rules
solidity-security	Reentrancy, access control, flash loans, oracle manipulation, gas optimization
rust-patterns	Anchor/Solana account validation, PDA derivation, CPI security
web3-frontend	Wallet connection, tx state, BigNumber display, error translation
smart-contract-audit	Pre-deployment security checklist (manual invoke)
web3-foundry	forge/cast/anvil commands, testing, deployment
web3-hardhat	Config, testing, deployment, plugins
web3-privy	Auth SDK, embedded wallets, wagmi integration
web3-eip-reference	Top 50 EIPs inline + lookup instructions
web3-solana-simd	Key Solana improvement proposals + concepts
web3-monad	MonadBFT, parallel execution, deployment
web3-megaeth	Full MegaETH dev reference (14 files)
arbitrum-dapp-skill	Arbitrum dApp development with Stylus (Rust) and Solidity (7 files)
web3-polymarket	Polymarket CLOB integration: auth, orders, market data, WebSocket, CTF ops, bridge
web3-solidity-patterns	Factory, Proxy, Diamond, Governor patterns

Agents (3)

Agent	Purpose
codebase-search	Web3-aware codebase search (ABIs, selectors, storage, Anchor)
open-source-librarian	Find implementation details in OSS repos with GitHub permalinks
tech-docs-writer	Create accurate technical documentation

Rules (6)

Rule	Triggers On	Key Points
typescript.md	**/*.{ts,tsx}	bigint for tokens, Address type, receipt checks
solidity.md	**/*.sol	NatSpec, CEI, custom errors, events
rust.md	**/*.rs	No unwrap, thiserror/anyhow, iterators
forge.md	**/*.sol, foundry.toml	Multi-chain config (8 chains), deployment rules
testing.md	**/*.{test,spec}.ts	BDD comments, one assertion per test
comments.md	All code files	When comments are forbidden vs required

Commands (3)

Command	Usage
/audit [contract]	Structured smart contract security audit
/deploy-check [chain] [contract]	Pre-deployment verification checklist
/interview [plan-file]	Interactive spec/plan interview

Hooks (3)

Hook	Trigger	What It Does
keyword-detector.py	Every prompt	Detects 10 keyword modes, injects context
check-comments.py	After Write/Edit	Warns if comment ratio exceeds 25%
todo-enforcer.sh	On session stop	Blocks exit with incomplete todos

Credits

• Arbitrum dApp skill: hummusonrails/arbitrum-dapp-skill
• MegaETH skills: 0xBreadguy/megaeth-ai-developer-skills
• Vercel React best practices: vercel/react-best-practices
• Planning with files: Based on Manus context engineering
• React useEffect patterns: React official docs

License

MIT