1. Conduid
  2. Agents
  3. Web3 Claude
MCP server · Agents

Web3 Claude

Open-source sample of a production Claude Code agent framework — 100+ specialized subagents for coordinated parallel work in Web3/DeFi

34Low

Scored 6 months ago · breakdown

About Web3 Claude

Web3 Claude is an MCP server in the Agents category: open-source sample of a production Claude Code agent framework — 100+ specialized subagents for coordinated parallel work in Web3/DeFi.

Install

From source
git clone https://github.com/johnpphd/Web3-Claude

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Ask AI

✦

Ask AI about Web3 Claude

Powered by Claude · Grounded in docs

I know everything about Web3 Claude. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

  • ✓README presentThe repository ships documentation.

5 checks have not been run yet.

README

Web3-Claude Configuration Template

A comprehensive, reusable Claude Code configuration template for React + Web3 + React Query projects. This template provides specialized agents, workflows, and documentation to accelerate Web3 application development with AI assistance.

🚀 Quick Start

  1. Copy the .claude folder to your project root
  2. Follow the portability guide to customize for your project
  3. Run /skill-sync all to regenerate project-specific files

Using this template? See .claude/docs/portability-guide.md for detailed instructions on what's reusable, what needs parameterization, and what must be regenerated.

⚠️ Security Notice

Before using in production:

  • Never commit API keys, secrets, or private keys
  • Review .gitignore to ensure sensitive files are excluded
  • Read SECURITY.md for security best practices
  • Use environment variables for all sensitive configuration

📁 Repository Structure

File / Directory Purpose
.claude/CLAUDE.md Project instructions loaded automatically every conversation
.claude/settings.json Pre-allowed permissions to reduce prompts
.claude/hooks.json Automatic actions (currently: Prettier on file save)
.claude/commands/ Simple slash commands for common workflows
.claude/skills/ Specialized agents and complex workflow skills
.claude/docs/ Shared reference documents (single source of truth)

Architecture

Shared Reference Documents (docs/)

These are the single source of truth for project knowledge. Skills reference them instead of embedding duplicate content.

Document Contents
docs/project-rules.md All coding rules, safety patterns, anti-patterns
docs/component-reference.md Common component APIs and usage examples
docs/theme-reference.md Full palette and typography tables
docs/data-patterns.md Ponder query patterns, blockchain write templates
docs/portability-guide.md Guide for porting .claude to new projects
docs/PROTOCOL_SPECIFICATION.md Protocol-level entity design (project-specific)

Agent Hierarchy

agent-orchestrator .............. Top-level task routing
├── ui-designer ................ All UI changes, layout, visual design
│   ├── ui-design-specialist ... Anti-slop design critic (read-only)
│   ├── theme-ui-specialist .... Palette, typography, styled(), MUI overrides
│   └── react-specialist ....... Component logic, hooks, state, performance
├── web3-implementer ........... All blockchain + ponder data work
│   ├── wagmi-specialist ....... Contract reads/writes, tx lifecycle, viem
│   └── react-query-specialist . Cache strategy, query keys, invalidation
└── typescript-specialist ...... Advanced types, generics, type safety (shared)

Standalone Workflow Skills

These are user-invoked skills that span domains (not part of the agent hierarchy):

Skill Type Purpose
/analyze-theme Inline Theme compliance auditor
/verify-app Fork Comprehensive verification (typecheck, lint, build, security)
/code-simplifier Fork Post-implementation cleanup
/skill-sync Inline Sync .claude knowledge files with codebase
/new-component Inline Component scaffold following project conventions
/new-hook Inline Hook scaffold with layer-appropriate templates

Commands

Simple slash commands in commands/. Type /command-name to invoke.

Command Purpose
/verify Run typecheck, lint, prettier, build
/fix-lint Auto-fix linting and formatting
/fix-number-format Fix number formatting anti-patterns
/commit-push-pr Commit, push, and create PR
/update-contracts Update contract ABIs and addresses
/verify-ui UI verification checklist

Recommended Workflows

Standard Feature Development

  1. Start in Plan mode (Shift+Tab twice)
  2. Discuss and refine the plan with Claude
  3. Switch to normal mode and implement
  4. Run /verify to check for issues
  5. Run /code-simplifier to clean up
  6. Run /commit-push-pr to ship

Quick Fix

  1. Make the fix
  2. Run /fix-lint
  3. Run /commit-push-pr

Pre-Release Check

  1. Run /verify-app for comprehensive verification
  2. Address any issues found
  3. Run /commit-push-pr

Theme Audit

  1. Run /analyze-theme to find violations
  2. Review the report and choose fix scope
  3. Run /verify after fixes

Customization

Adding New Skills

Use /skills-creator to create new skills. It guides you through requirements gathering, design review, and agent hierarchy integration.

Adding New Commands

Create a .md file in commands/. The filename becomes the command name. Use commands for simple workflows (< 50 lines, no domain knowledge, no supporting files).

Updating Project Context

  • CLAUDE.md: Quick reference card, top rules, slash command table
  • docs/project-rules.md: All detailed coding rules and conventions
  • docs/component-reference.md: Common component API reference
  • docs/theme-reference.md: Palette and typography lookup tables

Hooks

Edit hooks.json to add automatic actions. Available hooks: PostToolUse, PreToolUse, Stop.


Porting to New Projects

This .claude folder is ~85% reusable for any React + wagmi + Ponder project with the same design system approach.

What's Universal (Copy As-Is)

  • All QA skills (visual-qa, accessibility, responsive, performance)
  • All design philosophy skills (ui-design-specialist, ui-design-jony-ive, design-dialogue)
  • All refactoring specialists
  • Core workflow skills (verify, code-simplifier, skills-creator)
  • Permission settings (settings.json, hooks.json)

What Needs Entity Name Replacement

  • CLAUDE.md, README.md, project-rules.md
  • agent-orchestrator files
  • web3-implementer files
  • Hook pattern examples

Replace entity placeholders with your domain entities.

What Must Be Regenerated

Run /skill-sync all after copying the skill-sync folder — it handles all of these:

  • .claude/docs/theme-reference.md — from themeConfig.tsx
  • ponder-schema-specialist/schema-reference.md — from ponder.schema.ts
  • wagmi-specialist/hook-reference.md — from src/hooks/blockchain/
  • wagmi-specialist/contracts-reference.md — from generated.ts
  • web3-implementer/ponder-reference.md — from src/hooks/ponder/
  • typescript-specialist/type-index.json — from src/types/
  • skills/routes.json — from router config + src/pages/

Quick Start: Setup Prompt

After copying .claude to a new project, invoke /skills-creator setup and tell it about your project:

New protocol (different entities):

/skills-creator setup

I copied this .claude folder from another project. This is a new lending
protocol called "Aqueduct". My primary entity is "pool" and secondary
entity is "aggregator". Same tech stack: React + wagmi + Ponder + MUI.

Same protocol family (shared indexer/contracts):

/skills-creator setup

I copied this .claude folder from a sibling app. This project shares the
same ponder.schema.ts and generated.ts (same indexer and contracts). Same
entities, just different theme and pages.

Full Guide

See .claude/docs/portability-guide.md for:

  • Complete file-by-file classification
  • Step-by-step setup instructions
  • Same-family setup (shared indexer/contracts)
  • Architectural patterns reference
  • Setup checklist

🤝 Contributing

Contributions are welcome! Please read CONTRIBUTING.md for details on how to contribute to this project.

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🔒 Security

Please review our Security Policy before using this template in production. Never commit secrets, API keys, or private keys to your repository.

💡 Support


Note: This is a configuration template for Claude Code. It does not contain application code, but rather AI agent configurations and workflows to assist in Web3 development.

README copy captured when this server was indexed; it may lag the repository. The original is authoritative.

Questions

About Web3 Claude

How do I install Web3 Claude?

No package has been published for this server, so install it from source: clone the repository and follow its README.

Is Web3 Claude safe to use with an AI agent?

Its trust score is 34 out of 100 (low). It passes 1 of the 1 static security checks run so far. It has no ConduID identity yet, so agent calls to it are not receipted.

Is Web3 Claude still maintained?

Conduid hasn't recorded a commit date for this repository yet. Check the repository directly for recent activity.