Comprehensive collection of 76+ specialized agent prompts for Claude Code Organized, categorized, and ready to integrate into any project
💝 Support this project: GMr9rXrFSt5H3xX1wi85vyCQfwwLsQpULwNEjrSghvRN (View on Solscan)
📖 Quick Start | 🔧 Integration Guide | 🤖 Automation | 📝 Changelog | 🚀 Releases
This library provides a curated collection of specialized agent prompts extracted from the claude-flow project. Each prompt is carefully organized by category and designed to work seamlessly with Claude Code and MCP (Model Context Protocol).
- 76+ Specialized Agent Prompts - Covering all aspects of software development
- 9 Main Categories - Core, GitHub, Hive-Mind, Swarm, SPARC, Optimization, Testing, DevOps, Analysis
- Interactive CLI Tool - Browse, search, and integrate with ease
- Adaptive Automation - 3 automation levels to match your DevOps workflow ⭐ NEW
- Setup Wizard - Auto-detects your environment and recommends best automation
- Quick Presets - Pre-configured bundles for common workflows
- Complete Documentation - Each agent includes detailed instructions
Adaptive Automation System - Choose the perfect automation level for your workflow:
- Full Automation (GitHub Actions) - Zero manual work, automatic everything
- Smart Wizard (Platform Agnostic) - Intelligent tools for any CI/CD platform
- Lite Templates (Manual Control) - Lightweight templates with maximum flexibility
Run the setup wizard to get started:
cd agent-prompts
./setup-wizard.js👉 Read the complete automation guide
agent-prompts/
├── core/ # Essential development agents (coder, planner, tester, reviewer)
├── github/ # GitHub integration (PR manager, code review, issues)
├── hive-mind/ # Distributed intelligence coordination
├── swarm/ # Advanced swarm patterns
├── sparc/ # SPARC methodology (5-phase development)
├── optimization/ # Performance and benchmarking
├── testing/ # Comprehensive testing strategies
├── devops/ # CI/CD and automation
└── analysis/ # Code quality analysis
Beautiful CLI interface with:
- 🔍 Search - Find agents by name, capability, or use case
- 📚 Browse - Explore by category with breadcrumbs
- ✓ Select - Multi-select agents for integration
- 🚀 Integrate - One-click integration into your project
- ⚡ Presets - Quick bundles for common workflows
Three ways to integrate:
-
Interactive CLI (Recommended)
npm run integrate # Or directly: ./integrate.js -
Quick Script
./quick-integrate.sh
-
Manual Copy
cp -r agent-prompts/core .claude/agents/
- Node.js 18+ (for interactive CLI)
- Claude Code
- Git (optional, for GitHub agents)
# 1. Navigate to your project
cd your-project
# 2. Clone or copy this library
cp -r /path/to/agent-prompts ./
# 3. Run the interactive integration tool
cd agent-prompts
npm install # First time only
npm run integrate
# 4. Follow the interactive prompts to select and integrate agentsThe interactive CLI provides the easiest way to browse and integrate agents:
./integrate.jsFeatures:
- Browse all 9 categories
- Search by keyword
- View detailed agent information
- Select individual or bulk agents
- Generate integration scripts
- Use quick presets
Pre-configured bundles for common use cases:
Agents: coder, tester, reviewer, planner
Use for: Complete web application development
Agents: pr-manager, code-review-swarm, issue-tracker, ci-cd-github
Use for: PR management, code review, CI/CD automation
Agents: tester, reviewer, code-analyzer, performance-monitor
Use for: Testing, analysis, optimization
Agents: specification, pseudocode, architecture, refinement
Use for: Structured 5-phase development
Agents: queen-coordinator, collective-intelligence, hierarchical-coordinator
Use for: Multi-agent coordination
Essential agents for everyday development tasks.
| Agent | Description | Priority | Use Case |
|---|---|---|---|
| coder | Implementation specialist | High | Write clean, efficient code |
| planner | Task planning & orchestration | High | Break down complex tasks |
| researcher | Requirements analysis | Medium | Understand requirements |
| reviewer | Code quality & security | High | Review code, audit security |
| tester | Test generation & validation | High | Write tests, practice TDD |
Specialized agents for GitHub workflows and automation.
| Agent | Description | Priority | Use Case |
|---|---|---|---|
| pr-manager | Comprehensive PR management | High | Manage pull requests |
| code-review-swarm | Multi-agent code review | High | Parallel code reviews |
| issue-tracker | Issue management & triage | Medium | Organize issues |
| release-manager | Release automation | Medium | Automate releases |
Distributed intelligence and swarm coordination.
| Agent | Description | Priority | Use Case |
|---|---|---|---|
| queen-coordinator | Sovereign orchestrator | Critical | Top-level coordination |
| collective-intelligence | Distributed decision-making | High | Consensus building |
| scout-explorer | Reconnaissance specialist | Medium | Explore codebases |
| worker-specialist | Task execution | Medium | Execute specific tasks |
Advanced coordination patterns and topologies.
| Agent | Description | Priority | Use Case |
|---|---|---|---|
| hierarchical-coordinator | Tiered command structure | High | Structured coordination |
| mesh-coordinator | Peer-to-peer network | High | Decentralized control |
| adaptive-coordinator | Self-organizing topology | High | Dynamic adaptation |
5-phase structured development approach.
| Phase | Agent | Description | Use Case |
|---|---|---|---|
| 1 | specification | Requirements analysis | Define specifications |
| 2 | pseudocode | Algorithm design | Design before coding |
| 3 | architecture | System design | Plan architecture |
| 4 | refinement | Code optimization | Refine implementation |
| 5 | completion | Integration | Complete the cycle |
Performance monitoring and optimization.
| Agent | Description | Priority | Use Case |
|---|---|---|---|
| performance-monitor | Real-time monitoring | Medium | Track performance |
| benchmark-suite | Benchmarking framework | Medium | Compare performance |
Comprehensive testing strategies.
| Agent | Description | Priority | Use Case |
|---|---|---|---|
| tdd-london-swarm | TDD coordination | High | Practice TDD |
| production-validator | Production validation | High | Validate deployments |
DevOps automation and pipelines.
| Agent | Description | Priority | Use Case |
|---|---|---|---|
| ci-cd-github | GitHub Actions automation | High | Setup CI/CD |
Code quality and analysis.
| Agent | Description | Priority | Use Case |
|---|---|---|---|
| code-analyzer | Quality & complexity analysis | Medium | Analyze code quality |
./integrate.jsAdvantages:
- Visual browsing with categories
- Search functionality
- Preview prompt content
- Generate integration scripts
- Use quick presets
For automated or CI/CD integration:
./quick-integrate.sh [preset-name]Presets:
fullstack- Full-stack developmentgithub- GitHub workflowquality- Code quality focussparc- SPARC methodologyswarm- Swarm intelligenceall- All agents
Example:
./quick-integrate.sh github
# Integrates: pr-manager, code-review-swarm, issue-tracker, ci-cd-githubCopy specific categories:
# Core development agents
cp -r agent-prompts/core .claude/agents/
# GitHub integration
cp -r agent-prompts/github .claude/agents/
# SPARC methodology
cp -r agent-prompts/sparc .claude/agents/Generate a custom shell script:
./integrate.js
# Select agents interactively
# Choose "Generate integration script"
# Run the generated script
./integrate-agents.shOnce integrated, use agents in your prompts:
Use the coder agent to implement user authentication
Coordinate with pr-manager to review and merge PR #42
Follow the SPARC methodology:
1. specification - Define requirements
2. pseudocode - Design algorithm
3. architecture - Plan structure
4. refinement - Implement with TDD
5. completion - Integrate and deploy
Use hive-mind coordination:
- queen-coordinator: Oversee the task
- scout-explorer: Analyze the codebase
- worker-specialist: Implement features
- collective-intelligence: Build consensus
Each agent prompt is a markdown file with YAML frontmatter:
---
name: coder
type: developer
color: "#FF6B35"
description: Implementation specialist
capabilities:
- code_generation
- refactoring
priority: high
---
# Agent Instructions
...Edit any prompt to customize behavior, add tools, or modify instructions.
- Create a new
.mdfile in the appropriate category - Add YAML frontmatter with metadata
- Write agent instructions in markdown
- Update
catalog.jsonto include the new agent
Edit integrate.js and add to the presets array:
{
name: 'My Custom Preset',
icon: '🎯',
description: 'Description of the preset',
agents: ['agent1', 'agent2', 'agent3']
}- Total Agents: 76+
- Categories: 9 main categories
- Total Lines: 27,000+ lines of prompts
- Source: Based on claude-flow v2.7.34
- License: Based on original source
# Initialize Claude Code
mkdir -p .claude/agents
# Integrate agents
cd /path/to/agent-prompts
./integrate.js
# Select agents or use presets
# They will be copied to your .claude/agents/ directory# Navigate to your project
cd your-existing-project
# Integrate agents (they'll merge with existing .claude/ structure)
/path/to/agent-prompts/integrate.js
# Select integration target: .claude/agents/Add to your CI pipeline:
# .github/workflows/setup.yml
- name: Integrate Claude Agents
run: |
./agent-prompts/quick-integrate.sh githubMany agents use MCP (Model Context Protocol) tools:
// Example from pr-manager.md
mcp__claude-flow__swarm_init { topology: "mesh", maxAgents: 4 }
mcp__claude-flow__agent_spawn { type: "reviewer" }
mcp__github__create_pull_request { ... }Coordinate multiple agents:
// Initialize swarm
mcp__claude-flow__swarm_init { topology: "hierarchical" }
// Spawn agents
mcp__claude-flow__agent_spawn { type: "coder" }
mcp__claude-flow__agent_spawn { type: "tester" }
mcp__claude-flow__agent_spawn { type: "reviewer" }
// Orchestrate task
mcp__claude-flow__task_orchestrate {
task: "Implement feature X",
strategy: "parallel"
}Share state between agents:
// Store in shared memory
mcp__claude-flow__memory_usage {
action: "store",
key: "swarm/shared/context",
value: { feature: "auth", status: "implementing" }
}
// Retrieve from memory
mcp__claude-flow__memory_usage {
action: "retrieve",
key: "swarm/shared/context"
}To contribute new agents or improvements:
- Fork this repository
- Add your agent in the appropriate category
- Update
catalog.jsonwith metadata - Test with the integration tool
- Submit a pull request
- Source Project: claude-flow
- Claude Code Docs: docs.anthropic.com
- MCP Documentation: Model Context Protocol guides
- Changelog: CHANGELOG.md
- Releases: RELEASES.md
Release Date: November 17, 2025
Status: Stable ✅
What's New in 1.1.0:
- 🤖 Adaptive Automation System - 3 automation levels (Full, Smart, Lite)
- 🧙 Setup Wizard - Auto-detects environment and recommends best option
- ⚡ Full Automation - GitHub Actions workflows for automatic version/changelog/PR comments
- 🛠️ Smart Wizard - Platform-agnostic tools with diagnostics and health monitoring
- 📋 Lite Templates - CI/CD templates for GitHub, GitLab, Jenkins, CircleCI
- 📚 AUTOMATION.md - Comprehensive automation guide with comparison matrix
Previous Releases:
- 1.0.0 (Nov 17, 2025) - Initial release with 76+ agent prompts and integration tools
See Also:
- Full Changelog - Detailed version history
- Releases - Official release notes
- Automation Guide - Automation options explained
- Migration Guide - Upgrade instructions
Based on claude-flow. Please refer to the original project for licensing information.
If you find this library helpful, consider supporting its development!
Solana Donations:
GMr9rXrFSt5H3xX1wi85vyCQfwwLsQpULwNEjrSghvRN
Your support helps maintain and improve this library. Thank you! 🙏
- Original Project: claude-flow by @ruvnet
- Organization & Integration: Fused Gaming
# 1. Install (first time only)
cd agent-prompts
npm install
# 2. Run interactive tool
./integrate.js
# 3. Select agents or use presets
# 4. Integrate into your project
# 5. Start using agents in Claude Code!Questions? Open an issue or check the documentation in each agent prompt file.
Ready to supercharge your Claude Code workflow? 🚀