Awesome Claude Code Subagents (Claude Skill) — Use Cases, Install & Live Demo

Why use it

Key features

100+ subagents across major dev domains
Each has scoped tools + system prompt
Designed to be called from a parent Claude Code session
Coverable: security review, performance, codegen review, infra cost, etc.
Maintained, growing, community contributions

Live Demo

What it looks like in practice

ready

Install

Pick your client

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "awesome-claude-subagents": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/VoltAgent/awesome-claude-code-subagents",
        "~/.claude/skills/awesome-subagents"
      ],
      "_inferred": true
    }
  }
}

Open Claude Desktop → Settings → Developer → Edit Config. Restart after saving.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "awesome-claude-subagents": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/VoltAgent/awesome-claude-code-subagents",
        "~/.claude/skills/awesome-subagents"
      ],
      "_inferred": true
    }
  }
}

Cursor uses the same mcpServers schema as Claude Desktop. Project config wins over global.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "awesome-claude-subagents": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/VoltAgent/awesome-claude-code-subagents",
        "~/.claude/skills/awesome-subagents"
      ],
      "_inferred": true
    }
  }
}

Click the MCP Servers icon in the Cline sidebar, then "Edit Configuration".

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "awesome-claude-subagents": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/VoltAgent/awesome-claude-code-subagents",
        "~/.claude/skills/awesome-subagents"
      ],
      "_inferred": true
    }
  }
}

Same shape as Claude Desktop. Restart Windsurf to pick up changes.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "awesome-claude-subagents",
      "command": "git",
      "args": [
        "clone",
        "https://github.com/VoltAgent/awesome-claude-code-subagents",
        "~/.claude/skills/awesome-subagents"
      ]
    }
  ]
}

Continue uses an array of server objects rather than a map.

~/.config/zed/settings.json

{
  "context_servers": {
    "awesome-claude-subagents": {
      "command": {
        "path": "git",
        "args": [
          "clone",
          "https://github.com/VoltAgent/awesome-claude-code-subagents",
          "~/.claude/skills/awesome-subagents"
        ]
      }
    }
  }
}

Add to context_servers. Zed hot-reloads on save.

claude mcp add awesome-claude-subagents -- git clone https://github.com/VoltAgent/awesome-claude-code-subagents ~/.claude/skills/awesome-subagents

One-liner. Verify with claude mcp list. Remove with claude mcp remove.

Use Cases

Real-world ways to use Awesome Claude Code Subagents

Delegate a security review to a specialist subagent

👤 Devs shipping features that touch auth/data ⏱ ~20 min intermediate

When to use: Your main agent wrote auth code; you want a second pair of eyes from a focused subagent.

Flow

Spawn

Use awesome-subagents. Spawn the 'security-reviewer' subagent on src/auth.✓ Copied

→ Subagent reports findings
Triage

Group findings by severity. Identify which the main agent should fix now vs file as issues.✓ Copied

→ Triaged list

Outcome: Security feedback on every PR without paying a real reviewer's salary.

Pitfalls

Subagent finds things the main agent disagrees with — Your call — usually the security-focused output is right; default to caution

Combine with: claude-hud-skill

Run a performance audit subagent over a hot path

👤 Devs profiling slow features ⏱ ~45 min intermediate

When to use: An endpoint is slow and you want focused performance review.

Flow

Scope

Spawn the 'perf-auditor' subagent on the /orders endpoint. Include the DB schema.✓ Copied

→ Findings: N+1, missing indexes, hot loops
Apply

Apply the top 3 fixes; benchmark before/after.✓ Copied

→ Latency drop with numbers

Outcome: Real performance gains, not vibes.

Pitfalls

Subagent suggests micro-optimizations — Push for impact ranking; ignore noise

Find the right subagent for a niche task

👤 Devs new to the catalog ⏱ ~15 min beginner

When to use: You have a specific task and don't know which subagent to use.

Flow

Search

List subagents that are good for 'Postgres query optimization'.✓ Copied

→ 2-3 candidates with scope
Try

Use the top match on /reports/slow_queries.sql.✓ Copied

→ Targeted analysis

Outcome: Right tool for the job; less generic agent reasoning.

Pitfalls

Subagent definitions overlap — Pick the most narrowly scoped one

Combinations

Pair with other MCPs for X10 leverage

awesome-claude-subagents + claude-hud-skill

Spawn subagents and watch their progress live

Spawn 3 reviewers; track via HUD.✓ Copied

awesome-claude-subagents + mcp-agent-mail

Subagents coordinate via mail to avoid duplicate work

Subagents post claims to #review thread before starting.✓ Copied

Tools

What this MCP exposes

Tool	Inputs	When to call	Cost
list_subagents	domain?	Discovery	free
spawn_subagent	name, scope, task	Delegate a focused task	tokens for the subagent run

Cost & Limits

What this costs to run

API quota: N/A
Tokens per call: Each subagent run is its own context — costs add up; budget consciously
Monetary: Free
Tip: Use subagents only when the task is truly specialist; don't over-delegate

Security

Permissions, secrets, blast radius

Minimum scopes: Per-subagent — some need shell, some don't

Credential storage: None inherent

Data egress: Depends on subagent's tools

Never grant: Wide tool access to subagents that don't need it

Read each subagent's allowlist before granting

Troubleshooting

Common errors and fixes

Subagent runs forever

Set timeouts; some are over-eager

Subagent ignored scope

Be more explicit in the spawn task; set strict path scope

Alternatives

Awesome Claude Code Subagents vs others

Alternative	When to use it instead	Tradeoff
Single multi-tool agent	Tasks are general-purpose	Less focused output
wshobson/agents	You want a different curated set	Overlap; pick the philosophy you prefer

More

Resources

📖 Read the official README on GitHub

🐙 Browse open issues

🔍 Browse all 400+ MCP servers and Skills