Skip to content
Agentic Levels
Start hereLevelsTracksPromptsAbout
Get my levelGet my level

  • Start hereYour first useful task with AI.
  • Level quiz8 questions. No AI needed.
  • Fluency ScoreYour Level and a score, from your own AI.

  • LevelsThe 11 levels, in order.
  • LessonsShort lessons, filter by level.
  • TracksGuided paths by tool, role or goal.
  • Set up your AI tool
  • Starter packFree files that help AI remember.

  • PromptsReady prompts to copy and adapt.
  • ChatGPT vs Claude vs GeminiPick the right one for the job.
  • How-tos
  • ReferenceCommands for Claude Code, Codex, Antigravity.
  • ResourcesHand-picked reading and tools.

  • Your progress
  • Your next moveAn interview that picks your next skill.
  • Your next lessonsThe next lessons at your level.
  • Updates

  • Pricing
  • About
  • FAQ & Feedback
  • Work with us
  • How we handle your data
Sign inPreferences
AboutPricingHow-tosPromptsStarter packMy stackHow we handle your data

© 2026 Fuentes Studio

PrivacyTermsRefundsAccessibilityAI content policyCredits
yourCouncil
Ready to help
✦

What do you want to understand?

Ask anything about what you're learning.

←How-to guides
  1. Resources
  2. ›How-to guides

How to use Awesome Claude Code Subagents

131+ drop-in Claude Code subagents across 10 categories, with intelligent Opus/Sonnet/Haiku routing and granular tool permissions per role.

Levels 4 to 6On GitHub

Building a subagent from scratch means writing the system prompt, defining tools, deciding scope, and testing behavior. The awesome-claude-code-subagents collection gives you 131+ pre-configured agents across 10 categories: core development, language specialists, infrastructure, security, data/AI, fintech, healthcare, blockchain, and more. Intelligent model routing automatically assigns Opus for complex architectural tasks, Sonnet for standard work, and Haiku for simple operations. Granular tool permissions prevent agents from accessing tools outside their scope.

Before you start

Claude Code at Level 4 and at least one agent or tool-calling session under your belt. The subagent configs are markdown files, no additional installs required.

  1. 1Browse by agent role
  2. 2Copy an agent config into your project
  3. 3Invoke the agent with a scoped task
  4. 4Add project context to the agent's markdown file
  5. 5Use the parallel research pattern to dispatch subagents

Steps

01

Browse by agent role

The collection is organized by role: frontend, backend, security, devops, data, and others. Each agent has a README describing its scope, what it handles well, and what it explicitly does not do. Read the 'Does not do' section. It tells you when to use a different agent.

02

Copy an agent config into your project

Each agent is a single markdown file. Copy it into `.claude/agents/` in your project root. That is the only installation step. Claude Code discovers it there at session start.

Copy the agent
mkdir -p [YOUR_PROJECT_ROOT]/.claude/agents cp [PATH_TO_REPO]/[AGENT_FOLDER]/[AGENT_FILE].md [YOUR_PROJECT_ROOT]/.claude/agents/
[YOUR_PROJECT_ROOT]Your project root directory e.g. /Users/me/projects/my-app
[PATH_TO_REPO]Where you cloned the collection e.g. /Users/me/code/awesome-claude-code-subagents
[AGENT_FOLDER]The agent's folder in the collection e.g. security-reviewer
[AGENT_FILE]The agent's markdown filename e.g. security-reviewer
03

Invoke the agent with a scoped task

In a Claude Code session, call the agent by name and give it a narrow task, one file or one feature, not the whole codebase. Subagents produce generic output when given broad inputs.

Invoke the agent
@[AGENT_NAME] [SPECIFIC_TASK]
[AGENT_NAME]Agent name from the config file e.g. security-reviewer
[SPECIFIC_TASK]Scoped task for the agent e.g. Review src/lib/auth.ts for SQL injection, insecure token storage, and missing rate limiting
04

Add project context to the agent's markdown file

Open the agent's file and add a short section at the bottom: your stack, your conventions, and assumptions the agent should make. A security reviewer that knows you use Supabase RLS will give you more relevant findings than a generic one.

Add to the agent file
## Project Context Stack: [YOUR_STACK] Conventions: [YOUR_KEY_CONVENTIONS] Assume: [ASSUMPTIONS_FOR_THE_AGENT]
[YOUR_STACK]Your tech stack e.g. Next.js 15, Supabase, Anthropic SDK, deployed on Vercel
[YOUR_KEY_CONVENTIONS]Conventions the agent should respect e.g. all auth uses Supabase RLS, never raw SQL from the client
[ASSUMPTIONS_FOR_THE_AGENT]Baseline assumptions e.g. users are authenticated via Supabase Auth, JWTs validated server-side
05

Use the parallel research pattern to dispatch subagents

The collection's strongest move is parallel research: you name several targets, Claude spawns one subagent per target, then merges findings into a comparison. The prompt below fixes four output points per competitor so the parallel work converges into a scannable table, not five disconnected reports.

Parallel competitor research via subagents
Research these five competitors: [list them]. Look into each one in parallel, then bring the findings back together. For each competitor, return the same four points: what they do, who they are for, their pricing if it is public, and one thing they do better than the others. End with a short table comparing all five side by side so the differences are easy to scan.

How do you know it worked?

  • The agent markdown file is in `.claude/agents/` and Claude Code discovers it.
  • You invoked the agent on a real file and received output specific to your code.
  • The output reflects the agent's specialization, a security reviewer surfaced security concerns, not style feedback.
  • You added project context and the output improved on the second invocation.

If something looks off

Installing every agent from the collection
20 active agents means Claude Code has to reason about which applies to every task. Start with the one or two that address your most frequent needs.
Using a subagent for general tasks
The default Claude Code session is already a capable general agent. Subagents add value only when you need a specific, consistent viewpoint: security, performance, database design.
Agent configs going stale when your stack changes
If you change your auth provider or add a framework, update the project context section in affected agents. A stale context section actively misleads the agent.

Skills & Commands Reference

Subagent lifecycle, copy, invoke, and customize from the collection.

Command / SkillWhat it doesWhen to use it
.claude/agents/Drop any agent markdown file here, Claude Code discovers it at session start with no other install stepInstalling any agent from the collection
@[agent-name] [task]Invoke a specific agent for a scoped task in a Claude Code sessionNarrow, domain-specific work: security review, DB schema, API design, performance audit
Agent READMEEach agent's 'Does not do' section lists explicit scope limitsRead before invoking, tells you when this agent is the wrong choice
Project Context sectionAppend your stack, conventions, and baseline assumptions to the agent's markdown fileFirst thing after copying, a generic agent without stack context gives generic output
Scope a task tightlyPass one file or one feature to a subagent, not the whole codebaseEvery invocation, broad inputs produce broad, unfocused findings

See also

  • Anthropic Skills Library: official skills to pair with your subagents

What you'll do, 5 steps

  1. 1Browse by agent role
  2. 2Copy an agent config into your project
  3. 3Invoke the agent with a scoped task
  4. 4Add project context to the agent's markdown file
  5. 5Use the parallel research pattern to dispatch subagents

Was this helpful?