Web3 Claude
Open-source sample of a production Claude Code agent framework β 100+ specialized subagents for coordinated parallel work in Web3/DeFi
Ask AI about Web3 Claude
Powered by Claude Β· Grounded in docs
I know everything about Web3 Claude. Ask me about installation, configuration, usage, or troubleshooting.
0/500
Reviews
Documentation
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.
π Quick Start
- Copy the
.claudefolder to your project root - Follow the portability guide to customize for your project
- Run
/skill-sync allto 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.
β οΈ Security Notice
Before using in production:
- Never commit API keys, secrets, or private keys
- Review
.gitignoreto ensure sensitive files are excluded - Read SECURITY.md for security best practices
- Use environment variables for all sensitive configuration
π 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
βββ ui-designer ................ All UI changes, layout, visual design
β βββ ui-design-specialist ... Anti-slop design critic (read-only)
β βββ theme-ui-specialist .... Palette, typography, styled(), MUI overrides
β βββ react-specialist ....... Component logic, hooks, state, performance
βββ web3-implementer ........... All blockchain + ponder data work
β βββ wagmi-specialist ....... Contract reads/writes, tx lifecycle, viem
β βββ react-query-specialist . Cache strategy, query keys, invalidation
βββ 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
- Start in Plan mode (Shift+Tab twice)
- Discuss and refine the plan with Claude
- Switch to normal mode and implement
- Run
/verifyto check for issues - Run
/code-simplifierto clean up - Run
/commit-push-prto ship
Quick Fix
- Make the fix
- Run
/fix-lint - Run
/commit-push-pr
Pre-Release Check
- Run
/verify-appfor comprehensive verification - Address any issues found
- Run
/commit-push-pr
Theme Audit
- Run
/analyze-themeto find violations - Review the report and choose fix scope
- Run
/verifyafter 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 β it handles all of these:
.claude/docs/theme-reference.mdβ fromthemeConfig.tsxponder-schema-specialist/schema-reference.mdβ fromponder.schema.tswagmi-specialist/hook-reference.mdβ fromsrc/hooks/blockchain/wagmi-specialist/contracts-reference.mdβ fromgenerated.tsweb3-implementer/ponder-reference.mdβ fromsrc/hooks/ponder/typescript-specialist/type-index.jsonβ fromsrc/types/skills/routes.jsonβ 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
π€ Contributing
Contributions are welcome! Please read CONTRIBUTING.md for details on how to contribute to this project.
π License
This project is licensed under the MIT License - see the LICENSE file for details.
π Security
Please review our Security Policy before using this template in production. Never commit secrets, API keys, or private keys to your repository.
π‘ Support
- π Read the documentation
- π Report bugs via GitHub Issues
- π¬ Share your experience and improvements
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.
