Skip to content
/
Tier A A6 Multi-agent orchestrator

VoltAgent awesome-claude-code-subagents

voltagent-subagents · VoltAgent/awesome-claude-code-subagents · ★ 21k · last commit 2026-05-25

Primitive shape 154 total
Subagents 154
00

Summary

VoltAgent awesome-claude-code-subagents — Summary

VoltAgent awesome-claude-code-subagents is a curated catalog of 154+ Claude Code subagents across 10 domain categories, installable as Claude Code plugins (voltagent-core-dev, voltagent-lang, voltagent-infra, etc.) or via manual copy / interactive installer script. Each subagent is a persona-md markdown file with YAML frontmatter specifying name, description, tools, and model. The organizing principle is domain taxonomy: language specialists (32 agents covering TypeScript, SQL, Swift, Vue, Angular, C++, Go, Rust, and many more), infrastructure (18), quality/security (19), data/AI (15), developer experience (17), specialized domains (16), business/product (18), meta-orchestration (13), and research/analysis (13). The 09-meta-orchestration category includes an agent-installer agent that can browse and install other agents from the repo itself — a self-referential pattern. VoltAgent (a JS/TS agent framework company) maintains this as an open-source contribution alongside their commercial product. Compared to seeds, this repo is most similar to BMAD-METHOD in its persona-md approach but is a pure catalog with no workflow orchestration layer, no skills, no commands, and no hooks — purely agents for Claude Code's native subagent mechanism.

01

Overview

VoltAgent awesome-claude-code-subagents — Overview

Origin

Created and maintained by VoltAgent (an agent framework company building VoltAgent JS/TS framework). 20,554 stars, last push 2026-05-25. MIT license. Active Discord community.

Philosophy

"This repository serves as the definitive collection of Claude Code subagents, specialized AI assistants designed for specific development tasks."

VoltAgent frames subagents as:

  • Independent context windows: Each operates in isolated context, preventing cross-contamination
  • Domain-specific intelligence: Carefully crafted instructions tailored to area of expertise
  • Shared across projects: Team-wide sharing ensures uniform development practices
  • Granular tool permissions: Each subagent can be configured with specific tool access rights

Catalog Philosophy

The README explicitly documents subagent advantages:

  • Memory Efficiency: Isolated contexts prevent main conversation clutter
  • Enhanced Accuracy: Specialized prompts lead to better results in specific domains
  • Workflow Consistency: Team-wide sharing ensures uniform approaches
  • Security Control: Tool access can be restricted per subagent type

Awesome List Format

Follows the "awesome list" GitHub convention — curated external links + in-repo agents. Category 09-meta-orchestration includes external links to third-party projects (airis-mcp-gateway, moai-adk, pied-piper, taskade) alongside in-repo agents.

Model Strategy

Each agent specifies model: sonnet or model: opus based on task complexity. No Haiku agents — all sonnet or opus.

02

Architecture

VoltAgent awesome-claude-code-subagents — Architecture

Distribution

  • Claude Code plugin marketplace (claude plugin marketplace add VoltAgent/awesome-claude-code-subagents)
  • Manual copy to ~/.claude/agents/ or .claude/agents/
  • Interactive installer script (install-agents.sh)
  • Standalone installer (no clone required, curl + bash)
# Plugin (recommended)
claude plugin marketplace add VoltAgent/awesome-claude-code-subagents
claude plugin install voltagent-lang

# Interactive installer
git clone https://github.com/VoltAgent/awesome-claude-code-subagents
cd awesome-claude-code-subagents
./install-agents.sh

# Standalone
curl -sO https://raw.githubusercontent.com/VoltAgent/awesome-claude-code-subagents/main/install-agents.sh
chmod +x install-agents.sh
./install-agents.sh

Directory Structure

categories/
  01-core-development/      # 12 agents (api-designer, backend-developer, etc.)
  02-language-specialists/  # 31 agents (typescript-pro, swift-expert, etc.)
  03-infrastructure/        # 17 agents (cloud-architect, kubernetes-specialist, etc.)
  04-quality-security/      # 18 agents (security-auditor, qa-expert, etc.)
  05-data-ai/               # 14 agents (ml-engineer, llm-architect, etc.)
  06-developer-experience/  # 16 agents (git-workflow-manager, dx-optimizer, etc.)
  07-specialized-domains/   # 15 agents (blockchain-developer, fintech-engineer, etc.)
  08-business-product/      # 17 agents (product-manager, technical-writer, etc.)
  09-meta-orchestration/    # 12 agents + external links
  10-research-analysis/     # 12 agents (research-analyst, market-researcher, etc.)
.claude-plugin/             # Plugin registry
.claude/                    # Claude Code integration files
CLAUDE.md
install-agents.sh
tools/                      # Utility scripts

Required Runtime

None beyond Claude Code.

Target AI Tools

Claude Code only (.claude/agents/ format).

Plugin Organization

Each category maps to one installable plugin:

  • voltagent-core-dev, voltagent-lang, voltagent-infra, voltagent-qa-sec
  • voltagent-data-ai, voltagent-dev-exp, voltagent-domains
  • voltagent-biz, voltagent-meta, voltagent-research
03

Components

VoltAgent awesome-claude-code-subagents — Components

Totals

Type Count
Subagents 154+
Commands 0
Skills 0
Hooks 0
MCP servers 0

Category Breakdown (154 in-repo agents + external links)

Category Count Plugin
01 Core Development 12 voltagent-core-dev
02 Language Specialists 31 voltagent-lang
03 Infrastructure 17 voltagent-infra
04 Quality & Security 18 voltagent-qa-sec
05 Data & AI 14 voltagent-data-ai
06 Developer Experience 16 voltagent-dev-exp
07 Specialized Domains 15 voltagent-domains
08 Business & Product 17 voltagent-biz
09 Meta & Orchestration 12 + external voltagent-meta
10 Research & Analysis 12 voltagent-research

Representative Agents Per Category

01 Core Development

  • api-designer — REST and GraphQL API architect
  • backend-developer — Server-side expert (Node 18+, Python 3.11+, Go 1.21+)
  • frontend-developer — UI/UX specialist (React, Vue, Angular)
  • fullstack-developer — End-to-end feature development
  • microservices-architect — Distributed systems designer
  • ui-designer — Visual design and interaction specialist
  • websocket-engineer — Real-time communication specialist

02 Language Specialists (31 agents)

  • typescript-pro, javascript-pro, python-pro
  • golang-pro, rust-engineer, java-architect
  • swift-expert, kotlin-specialist, csharp-developer
  • nextjs-developer, react-specialist, vue-expert
  • fastapi-developer, django-developer, rails-expert
  • laravel-specialist, symfony-specialist, flutter-expert
  • expo-react-native-expert, spring-boot-engineer
  • powershell-5.1-expert, powershell-7-expert

03 Infrastructure

  • kubernetes-specialist, terraform-engineer, cloud-architect
  • devops-engineer, sre-engineer, database-administrator
  • security-engineer, network-engineer, platform-engineer
  • docker-expert, deployment-engineer, incident-responder
  • terragrunt-expert, azure-infra-engineer, windows-infra-admin

04 Quality & Security

  • security-auditor, penetration-tester, compliance-auditor
  • code-reviewer, qa-expert, test-automator
  • performance-engineer, debugger, chaos-engineer
  • ai-writing-auditor, gdpr-ccpa-compliance
  • ad-security-reviewer, powershell-security-hardening
  • ui-ux-tester — exhaustive documented-flow UI tester

05 Data & AI

  • llm-architect, ml-engineer, mlops-engineer
  • data-engineer, data-scientist, data-analyst
  • postgres-pro, database-optimizer
  • reinforcement-learning-engineer, nlp-engineer
  • prompt-engineer

06 Developer Experience

  • git-workflow-manager, refactoring-specialist, legacy-modernizer
  • documentation-engineer, build-engineer, dx-optimizer
  • mcp-developer, cli-developer, slack-expert
  • powershell-module-architect, powershell-ui-architect
  • visual-asset-generator — 30+ image models via prompt-to-asset MCP

07 Specialized Domains

  • blockchain-developer, fintech-engineer, iot-engineer
  • healthcare-admin (51 sub-agents covering healthcare verticals)
  • hipaa-compliance, m365-admin
  • game-developer, embedded-systems, quant-analyst
  • payment-integration, seo-specialist

08 Business & Product

  • product-manager, project-manager, business-analyst
  • technical-writer, scrum-master, ux-researcher
  • wordpress-master, legal-advisor, license-engineer
  • content-marketer, content-quality-editor
  • growth-loops, assumption-mapping, backlog-grooming

09 Meta & Orchestration

In-repo agents:

  • agent-installer — browse/install agents from this repo via GitHub
  • agent-organizer — multi-agent coordinator
  • multi-agent-coordinator — advanced multi-agent orchestration
  • workflow-orchestrator — complex workflow automation
  • codebase-orchestrator — safe refactor governance
  • context-manager — context optimization

External links:

  • airis-mcp-gateway — Docker MCP multiplexer, 60+ tools behind 7 meta-tools (97% token reduction)
  • moai-adk — SPEC-first ADK, 24 agents, TRUST 5 quality gates
  • pied-piper — team-of-agents for SDLC workflows
  • taskade — AI workspace with MCP integration

10 Research & Analysis

  • research-analyst, competitive-analyst, market-researcher
  • data-researcher, scientific-literature-researcher
  • trend-analyst, first-principles-thinking
  • ab-test-analysis, cohort-analysis, project-idea-validator

Verbatim Agent Format

---
name: backend-developer
description: "Use this agent when building server-side APIs, microservices..."
tools: Read, Write, Edit, Bash, Glob, Grep
model: sonnet
---
05

Prompts

VoltAgent awesome-claude-code-subagents — Prompt Excerpts

Excerpt 1: backend-developer Agent

Technique: Checklist + API design requirements table (structured constraint encoding)

---
name: backend-developer
description: "Use this agent when building server-side APIs, microservices, and backend systems that require robust architecture, scalability planning, and production-ready implementation."
tools: Read, Write, Edit, Bash, Glob, Grep
model: sonnet
---

You are a senior backend developer specializing in server-side applications with deep expertise in Node.js 18+, Python 3.11+, and Go 1.21+.

When invoked:
1. Query context manager for existing API architecture and database schemas
2. Review current backend patterns and service dependencies
3. Analyze performance requirements and security constraints
4. Begin implementation following established backend standards

Backend development checklist:
- RESTful API design with proper HTTP semantics
- Database schema optimization and indexing
- Authentication and authorization implementation
- Caching strategy for performance
- Error handling and structured logging
- API documentation with OpenAPI spec
- Security measures following OWASP guidelines
- Test coverage exceeding 80%

API design requirements:
- Consistent endpoint naming conventions
- Proper HTTP status code usage
- Request/response validation
- API versioning strategy
- Rate limiting implementation
- CORS configuration
- Pagination for list endpoints
- Standardized error responses

Analysis: The "When invoked: 1. Query context manager" pattern is notable — the agent is instructed to check if a context-manager agent has prior context before starting. This implements soft cross-agent state passing via convention, not a formal protocol.

Excerpt 2: multi-agent-coordinator Agent (category 09)

Technique: Meta-orchestration with explicit coordination protocol documentation

This agent (full text not fetched but pattern from README) implements:

  • Agent selection based on task type
  • Parallel vs sequential coordination decisions
  • Result aggregation from multiple specialized agents

The agent-installer agent pattern:

"Browse and install agents from this repository via GitHub" "Find PHP agents and install php-pro globally"

This shows self-referential tooling: an agent whose purpose is to manage other agents in the same catalog.

09

Uniqueness

VoltAgent awesome-claude-code-subagents — Uniqueness & Positioning

Differs From Seeds

Closest to BMAD-METHOD (persona-md format, no commands/skills/hooks) but is a pure catalog without BMAD's orchestration layer or persona activation sequences. Unlike superpowers (14 skills, behavioral framework with Iron Laws), VoltAgent is pure agent definitions with no behavioral enforcement. The 10-category taxonomy is the most thorough domain coverage in this batch, including unusual verticals: healthcare-admin with 51 sub-agents, powershell-5.1-expert vs powershell-7-expert as distinct agents, and ad-security-reviewer (Active Directory). No seed approaches this category granularity. The sibling repo (voltagent-codex-subagents) providing TOML equivalents for Codex CLI represents the same multi-harness strategy as wshobson/agents but through dual-repo mirroring rather than an adapter framework.

Distinctive Position

Largest agent-only catalog with zero commands/skills/hooks. The healthcare-admin nesting (51 internal sub-agents in one agent file) is the most extreme domain specialization in this batch. The awesome-list format with external project links in category 09 extends the catalog beyond the repo itself.

Observable Failure Modes

  1. Catalog freshness: Popular frameworks/languages release new versions; agents specifying version numbers (Node.js 18+, Python 3.11+, Go 1.21+) will become stale without regular updates.
  2. Soft dependency conventions: The backend-developer agent saying "query context manager" assumes context-manager outputs exist — silent failure if not present.
  3. No quality gate: All 154+ agents are peer-contributed with no automated quality validation (unlike wshobson/agents' plugin-eval framework).
  4. Single-harness: Despite VoltAgent's sibling TOML repo, this repo requires migration to use with non-Claude-Code tools.

Explicit Antipatterns

None stated. README focuses on capabilities.

04

Workflow

VoltAgent awesome-claude-code-subagents — Workflow

User Workflow

  1. Install category plugins or individual agents
  2. Claude Code auto-spawns appropriate agent based on task context
  3. Or explicitly request: "Have the code-reviewer subagent analyze my latest commits"
  4. Agent executes in isolated context window

Agent Installer Pattern

The agent-installer agent within category 09 provides a self-referential workflow:

> Use the agent-installer to show me available categories
> Find PHP agents and install php-pro globally

This allows users to browse and install agents without leaving Claude Code.

Healthcare Admin Agent Pattern

The healthcare-admin agent in category 07 is documented as having 51 internal sub-agents covering revenue cycle, compliance, quality, clinical ops, health IT, and payer relations — an agent-within-agent pattern.

No Formal Phases

This is a pure catalog. No workflow phases, no approval gates, no spec-driven process. Installation → use.

Phase-to-Artifact Map

Phase Artifact
Install category plugin Agent files in ~/.claude/agents/
Claude Code task assignment Specialized agent response
Orchestration (meta category) Multi-agent coordinated output
06

Memory Context

VoltAgent awesome-claude-code-subagents — Memory & Context

State Storage

No persistent state. Each subagent operates in an isolated context window — this is explicitly documented as a feature ("Independent Context Windows: Every subagent operates within its own isolated context space, preventing cross-contamination between different tasks").

Soft Cross-Agent State Passing

The backend-developer agent instructs Claude to "Query context manager for existing API architecture" before starting — this is a convention-based soft dependency on the context-manager agent's outputs having been persisted to project files. Not a formal state protocol.

No Memory Mechanism

No memory files, no SQLite, no vector store. Pure file-based project context only.

Cross-Session Handoff

None. Agent files are the only persistence — they define behavior, not state.

07

Orchestration

VoltAgent awesome-claude-code-subagents — Orchestration

Multi-Agent Support

Yes — category 09 (voltagent-meta) is dedicated to orchestration agents: multi-agent-coordinator, workflow-orchestrator, agent-organizer, codebase-orchestrator.

Orchestration Pattern

Hierarchical (meta-orchestration agents coordinate domain specialists). No formal protocol — coordination is achieved through Claude Code's native Task tool.

Isolation Mechanism

Process-level (Claude Code's native subagent context isolation — explicitly the primary feature the README sells).

Multi-Model

No — all agents specify either model: sonnet or model: opus. No Haiku, no inherit, no multi-tier routing.

Execution Mode

Interactive-loop. Agents auto-activate based on task context or explicit user request.

Consensus

None.

Self-Referential Orchestration

The agent-installer agent in category 09 can browse this repo's own catalog and install agents — a self-referential tooling pattern where an agent manages agent catalog expansion.

Cross-Tool Portability

Low/single-tool. Agents use .md format in .claude/agents/ — Claude Code native. The sibling repo (voltagent-codex-subagents) provides TOML-format equivalents for Codex CLI.

08

Ui Cli Surface

VoltAgent awesome-claude-code-subagents — UI / CLI Surface

Dedicated CLI

None. install-agents.sh provides interactive category/agent selection.

Local Dashboard

None.

IDE Integration

Claude Code only (/agents command, .claude/agents/ directory).

Agent Discovery

/agents    # Claude Code's built-in agent manager

Or via agent-installer agent: "Use the agent-installer to show me available categories"

Installation Script

install-agents.sh provides interactive terminal UI: browse categories, select agents, install/uninstall.

Related frameworks

same archetype · same primary tool · same memory type

Claude-Flow / Ruflo ★ 55k
A6 Multi-agent orchestrator

Eliminates single-agent context limits and sequential bottlenecks by orchestrating fault-tolerant swarms of specialized AI agents…

Hermes Agent (NousResearch) ★ 168k
A6 Multi-agent orchestrator

Self-improving personal AI agent with closed learning loop, 7 terminal backends, and messaging gateway — not tied to any AI…

OpenCode ★ 165k
A6 Multi-agent orchestrator

Terminal-first AI coding agent with multi-model routing, native desktop app, and a typed .opencode/ configuration system for…

OpenHands ★ 75k
A6 Multi-agent orchestrator

Open-source AI software development platform (open-source Devin alternative) with Docker sandbox isolation, 77.6% SWE-bench…

DeerFlow ★ 70k
A6 Multi-agent orchestrator

Long-horizon superagent that researches, codes, and creates by orchestrating parallel sub-agents with isolated contexts in Docker…

oh-my-openagent (omo) ★ 60k
A6 Multi-agent orchestrator

Multi-provider AI agent orchestration for OpenCode: escape vendor lock-in by routing Sisyphus (Claude/Kimi/GLM) and Hephaestus…