Comprehensive agent library featuring 115+ specialized Opus 4.5 agents organized by domain, 55 development skills, and powerful exploration and code review commands for enhanced Claude Code workflows.
claude plugin install zircote/agentsAfter installing, verify the agents and commands are available:
# Test an agent (should return a helpful response)
claude "Using the python-pro agent, explain Python's GIL in one sentence"
# Verify agent count in Task tool
claude "List 3 available subagent types from the zircote plugin"You should see agents like python-pro, frontend-developer, code-reviewer in the Task tool's available subagents.
Specialized agents organized by domain in agents/:
| Category | Count | Directory | Examples |
|---|---|---|---|
| Core Development | 11 | 01-core-development/ |
frontend-developer, backend-developer, fullstack-developer, api-designer, mobile-developer, ui-designer |
| Language Specialists | 23 | 02-language-specialists/ |
python-pro, typescript-pro, golang-pro, rust-engineer, java-architect, react-specialist, nextjs-developer, vue-expert |
| Infrastructure | 12 | 03-infrastructure/ |
devops-engineer, sre-engineer, kubernetes-specialist, terraform-engineer, cloud-architect, database-administrator |
| Quality & Security | 12 | 04-quality-security/ |
code-reviewer, security-auditor, penetration-tester, test-automator, performance-engineer, debugger, qa-expert |
| Data & AI | 12 | 05-data-ai/ |
data-scientist, data-engineer, ml-engineer, llm-architect, postgres-pro, prompt-engineer, nlp-engineer |
| Developer Experience | 10 | 06-developer-experience/ |
documentation-engineer, cli-developer, build-engineer, refactoring-specialist, mcp-developer, dx-optimizer |
| Specialized Domains | 11 | 07-specialized-domains/ |
fintech-engineer, blockchain-developer, game-developer, iot-engineer, payment-integration |
| Business & Product | 10 | 08-business-product/ |
product-manager, technical-writer, ux-researcher, scrum-master |
| Meta Orchestration | 8 | 09-meta-orchestration/ |
multi-agent-coordinator, workflow-orchestrator, task-distributor |
| Research & Analysis | 6 | 10-research-analysis/ |
research-analyst, competitive-analyst, market-researcher, trend-analyst |
| Templates | 1 | templates/ |
Agent creation templates |
Development skills in skills/:
| Category | Skills |
|---|---|
| AI/Prompting | anthropic-prompt-engineer, anthropic-architect, claude-code, prompt-engineer |
| Frontend | frontend-development, frontend-design, aesthetic, ui-styling, canvas-design, web-artifacts-builder, web-frameworks, artifacts-builder |
| Backend | backend-development, databases, devops |
| Code Quality | code-review, debugging (5 sub-skills), problem-solving (6 sub-skills) |
| Tools & Utilities | mcp-builder, changelog-generator, skill-creator, engineer-skill-creator, template-skill, skill-share, repomix, sequential-thinking |
| Media | ai-multimodal, image-enhancer, video-downloader, slack-gif-creator, chrome-devtools, svg-graphics |
| Business | brand-guidelines, competitive-ads-extractor, content-research-writer, internal-comms, lead-research-assistant, shopify |
| Specialized | better-auth, datadog-entity-generator, invoice-organizer, file-organizer, python-deprecation-fixer, python-project-skel, raffle-winner-picker, theme-factory, webapp-testing |
Located in commands/:
| Command | Description |
|---|---|
/zircote:release |
Prepare and execute a plugin release with version bump, validation, and optional PR creation |
All agents follow Opus 4.5 best practices:
---
name: agent-name
description: >
[Role] specialist for [domain]. Use PROACTIVELY when [trigger conditions].
[Key capabilities and integration points].
model: inherit
tools: Read, Write, Bash, Glob, Grep, [additional-tools]
---
[Detailed agent instructions including protocols, execution patterns, and output formats]Agents are invoked via the Task tool with subagent_type:
Task(subagent_type="frontend-developer", prompt="Build a React component for user authentication")
Task(subagent_type="security-auditor", prompt="Review this codebase for vulnerabilities")
Task(subagent_type="python-pro", prompt="Refactor this module using best practices")
Deploy multiple specialists simultaneously for independent tasks:
# Parallel: Full-stack feature with security review
Task(subagent_type="frontend-developer", prompt="Build the UI")
Task(subagent_type="backend-developer", prompt="Build the API")
Task(subagent_type="security-auditor", prompt="Review for vulnerabilities")
# Sequential: Research then implement
research = Task(subagent_type="research-analyst", prompt="Research best practices for X")
# Then use findings to guide implementation
Task(subagent_type="python-pro", prompt="Implement based on research: {research}")
Reference agents in your project's CLAUDE.md for automatic routing:
## Agent Preferences
- Use `frontend-developer` for React/TypeScript work
- Use `python-pro` for Python implementation
- Use `code-reviewer` before merging PRsAgent templates in agents/templates/ provide starting points for creating new agents following Opus 4.5 conventions.
agents/
├── .claude-plugin/
│ └── plugin.json
├── agents/
│ ├── 01-core-development/
│ ├── 02-language-specialists/
│ ├── 03-infrastructure/
│ ├── 04-quality-security/
│ ├── 05-data-ai/
│ ├── 06-developer-experience/
│ ├── 07-specialized-domains/
│ ├── 08-business-product/
│ ├── 09-meta-orchestration/
│ ├── 10-research-analysis/
│ └── templates/
├── skills/
│ ├── aesthetic/
│ ├── ai-multimodal/
│ ├── anthropic-architect/
│ └── ... (55 total)
├── commands/
│ └── release.md
├── CHANGELOG.md
├── LICENSE
└── README.md
- Fork the repository
- Create a feature branch
- Make your changes
- Submit a pull request
VoltAgent/awesome-claude-code-subagents
MIT