
Specialized Claude Code sub-agents for modern full-stack development. These agents are designed to be used proactively when working on commits that involve their areas of expertise.
Expertise: Documentation maintenance and reference validation
Triggers: Agent configurations, documentation links, protocol changes
Purpose: Ensures all agent documentation remains current and accurate
Expertise: tRPC v11, TanStack Query, Prisma with PostgreSQL
Triggers: API endpoints, database schema, tRPC procedures, React Query, Prisma models
Purpose: Reviews and optimizes type-safe APIs with focus on performance and client hydration
Expertise: better-auth framework and security protocols
Triggers: Authentication flows, security configs, better-auth plugins, user management, sessions
Purpose: Ensures secure, user-friendly authentication implementations
Expertise: Turborepo and pnpm monorepo architecture
Triggers: turbo.json, pnpm-workspace.yaml, package.json, Dockerfiles, CI/CD workflows
Purpose: Maintains repository structure and CI integrity
Claude Code will automatically select the appropriate sub-agent based on your commits and task descriptions. Each agent above lists their trigger conditions.
-
Open the sub-agents interface:
/agents
-
Copy agents to your project:
- Copy the
.md
files from this repository to your project's.claude/agents/
directory - Or install them user-wide in
~/.claude/agents/
- Copy the
-
Explicit invocation:
> Use the api-agent to review my tRPC implementation > Use the auth-agent to check my better-auth configuration > Use the repo-agent to optimize my turbo.json setup
Each agent provides:
- β Review Checklist - Systematic validation of their domain
- π― Priority-based Feedback - CRITICAL/HIGH/MEDIUM issue categorization
- π§ Specific Fix Examples - Concrete code solutions for common problems
- π Proactive Triggers - Clear conditions for when they should be used
-
Install agents in your Claude Code project:
mkdir -p .claude/agents cp agents/*.md .claude/agents/
-
Use
/agents
in Claude Code to see available agents -
Make commits in their areas of expertise and watch them activate automatically!
PRs welcome for new agents or improvements to existing ones. Each agent should follow the established pattern with clear trigger conditions and actionable feedback.