Hive

110 Claude Code agent definitions organized into 22 specialized teams.

Teams

#	Team	Agents	Focus
1	Ops	5	System diagnostics, Docker, networking, hardware
2	Design	5	UI/UX, CSS, branding, marketing content
3	Full-Stack	5	Frontend, backend, APIs, data engineering
4	Extensions	4	Browser extension dev (Firefox/Chrome)
5	DevOps	4	Docker deployment, cloud platforms, CI/CD
6	Automation	4	CRM, Google Sheets, workflows, reports
7	QA & Security	4	Testing, security auditing, performance
8	Research	5	Codebase, web, system, network research
9	Compliance	5	Regulatory, privacy, policy, accessibility
10	Content	5	Technical writing, knowledge management
11	Finance	5	Cost analysis, pricing, ROI, business strategy
12	Incident Response	5	Incident management, runbooks, postmortems
13	AI & Prompt	5	Prompt engineering, AI features, agent architecture
14	Migration	5	Framework, schema, and platform migrations
15	UX Research	5	User research, analytics, usability testing
16	Marketing	6	SEO, social media, ads, lead gen, email/SMS
17	Sales	6	Sales process, proposals, coaching, CRM pipeline
18	Admin	5	Scheduling, email, documents, meetings
19	Accounting	5	Bookkeeping, commissions, payroll, taxes
20	Media Production	5	Video, graphics, presentations, infographics
21	Contracting	5	Agent contracting, credentialing, onboarding
22	AWS	7	CDK, DynamoDB, Lambda, API Gateway, IAM/Cognito, SES/SNS/SQS, CloudWatch

Structure

Each team has a lead (opus model) that coordinates specialists (sonnet model).

agents/
├── ops/                  # System operations
├── design/               # Visual design
├── fullstack/            # Web development
├── extensions/           # Browser extensions
├── devops/               # Deployment & infrastructure
├── automation/           # Business process automation
├── qa-security/          # Testing & security
├── research/             # Multi-domain research
├── compliance/           # Legal & regulatory
├── content/              # Writing & knowledge
├── finance/              # Financial analysis
├── incident-response/    # Reliability & incidents
├── ai-prompt/            # AI tooling optimization
├── migration/            # System migrations
├── ux-research/          # User experience research
├── marketing/            # Growth & marketing
├── sales/                # Sales operations
├── admin/                # Administrative support
├── accounting/           # Financial operations
├── media-production/     # Visual asset creation
├── contracting/          # Agent contracting & onboarding
└── aws/                  # AWS cloud services

Usage

These agents are designed for Claude Code subagents and agent teams.

Install globally

cp agents/**/*.md ~/.claude/agents/

Enable Agent Teams

Add to ~/.claude/settings.json:

{
  "env": {
    "CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS": "1"
  }
}

Using agents

As subagents — Claude automatically delegates to agents based on task context, or you can invoke them explicitly:

Use the platform-lead agent to diagnose my network issue
Have the security-auditor check this repo

As a team — Multiple agents work in parallel, communicate directly, and share a task list:

Create a team with fullstack-lead, qa-lead, and platform-lead to build and ship this feature

Each team has a lead (Opus) that coordinates and specialists (Sonnet) that execute. Leads create tasks, assign specialists, and synthesize results.

MCP Interns

Every specialist agent has access to two LLM "interns" via MCP servers — one powered by OpenAI and one by Google Gemini. These run as global MCP servers registered in ~/.claude.json.

Setup

# Install dependencies
cd mcp && python3 -m venv .venv
.venv/bin/pip install -r requirements.txt

Then configure auth in ~/.claude.json under the mcpServers env blocks:

• OpenAI: Set OPENAI_API_KEY — get a key at https://platform.openai.com/api-keys
• Gemini: Uses a GCP service account via Vertex AI. Set GOOGLE_SERVICE_ACCOUNT_KEY to the path of a service account JSON key file with Vertex AI access. The project defaults to woxom-sales-dashboard and location to us-east1 (override with GOOGLE_CLOUD_PROJECT and GOOGLE_CLOUD_LOCATION).

Available Tools

Both interns expose the same 5 tools:

Tool	Purpose
ask	General-purpose query (research, analysis, any free-form task)
draft	Write content (code, docs, emails, reports, proposals)
review	Critique content/code (security, writing, compliance, QA)
analyze	Structured analysis (cost-benefit, risk, competitive, root-cause)
brainstorm	Generate ideas and alternatives

Example

Use the openai-intern review tool to check this code for security issues
Have the gemini-intern draft a proposal for the new API design

Agent Format

Each .md file follows the Claude Code agent definition format:

---
name: agent-name
description: One-line description
model: opus | sonnet
---

[System prompt with role, capabilities, methodology, and rules]