Skip to content

TEMPLATE_CATALOG.md

Latest commit

 

History

History
568 lines (412 loc) · 12.2 KB

TEMPLATE_CATALOG.md

File metadata and controls

568 lines (412 loc) · 12.2 KB

Template Catalog

Complete reference for all 22 templates in the Claude PHP Agent framework.

Quick Reference

Category Templates Description
Agents 5 Basic and foundational patterns
Chatbots 2 Conversational agents with memory
RAG 3 Retrieval-augmented generation
Workflows 2 Multi-step and multi-agent systems
Specialized 5 Domain-specific agents
Production 2 Enterprise-ready configurations

Category: Agents

basic-agent.json

Difficulty: Beginner | Setup: 5 minutes | Icon: 🤖

A simple agent with one tool for basic tasks. Perfect for getting started with the framework.

Use Cases:

  • Learning the framework
  • Simple automation tasks
  • Single-tool operations

Configuration:

{
  "agent_type": "Agent",
  "model": "claude-sonnet-4-5",
  "max_iterations": 5
}

Example:

$agent = TemplateManager::instantiate('basic-agent', [
    'api_key' => getenv('ANTHROPIC_API_KEY'),
    'tools' => [$calculator]
]);

$result = $agent->run('What is 15 + 27?');

react-agent.json

Difficulty: Intermediate | Setup: 10 minutes | Icon: 🔄

Reason-Act-Observe pattern agent for general-purpose autonomous tasks. Uses iterative reasoning and tool execution.

Use Cases:

  • Research tasks
  • Data processing
  • Multi-step problem solving
  • Autonomous decision making

Configuration:

{
  "agent_type": "ReactAgent",
  "model": "claude-sonnet-4-5",
  "max_iterations": 10
}

Example:

$agent = TemplateManager::instantiate('react-agent', [
    'api_key' => getenv('ANTHROPIC_API_KEY'),
    'tools' => [$searchTool, $calculatorTool]
]);

$result = $agent->run('Research PHP 8.3 features and summarize them');

chain-of-thought-agent.json

Difficulty: Intermediate | Setup: 10 minutes | Icon: 🧠

Step-by-step reasoning agent that breaks down complex problems into logical steps for better accuracy.

Use Cases:

  • Complex problem solving
  • Mathematical reasoning
  • Logic puzzles
  • Analytical tasks

Example:

$agent = TemplateManager::instantiate('chain-of-thought-agent', [
    'api_key' => getenv('ANTHROPIC_API_KEY')
]);

$result = $agent->run('If 5 machines can make 5 widgets in 5 minutes, how long does it take 100 machines to make 100 widgets?');

reflex-agent.json

Difficulty: Beginner | Setup: 5 minutes | Icon: ⚡

Rule-based response agent for fast, deterministic reactions to specific conditions.

Use Cases:

  • FAQ responses
  • Simple automation
  • Conditional triggers
  • Pattern matching

model-based-agent.json

Difficulty: Intermediate | Setup: 15 minutes | Icon: 🗺️

State-aware decision making agent that maintains an internal model of the world for planning and prediction.

Use Cases:

  • Planning tasks
  • Simulation
  • State-based decision making
  • What-if analysis

Category: Advanced Agents

reflection-agent.json

Difficulty: Advanced | Setup: 15 minutes | Icon: 🪞

Self-improvement loop agent that generates output, reflects on quality, and refines until meeting quality threshold.

Use Cases:

  • Code generation
  • Content writing
  • Quality-critical outputs
  • Iterative improvement

Configuration:

{
  "agent_type": "ReflectionAgent",
  "max_refinements": 3,
  "quality_threshold": 8
}

Example:

$agent = TemplateManager::instantiate('reflection-agent', [
    'api_key' => getenv('ANTHROPIC_API_KEY')
]);

$result = $agent->run('Write a function to validate email addresses with comprehensive error handling');

plan-execute-agent.json

Difficulty: Advanced | Setup: 15 minutes | Icon: 📋

Multi-step planning agent that creates a plan first, then executes systematically for complex tasks.

Use Cases:

  • Project planning
  • Complex workflows
  • Multi-step tasks
  • Strategic execution

Example:

$agent = TemplateManager::instantiate('plan-execute-agent', [
    'api_key' => getenv('ANTHROPIC_API_KEY'),
    'allow_replan' => true
]);

$result = $agent->run('Create a complete marketing campaign for a new product launch');

tree-of-thoughts-agent.json

Difficulty: Advanced | Setup: 20 minutes | Icon: 🌳

Exploration agent that evaluates multiple reasoning paths simultaneously.

Use Cases:

  • Creative writing
  • Optimization problems
  • Strategy evaluation
  • Multi-path exploration

maker-agent.json

Difficulty: Advanced | Setup: 20 minutes | Icon: ⚙️

Million-step reliable task agent using massively decomposed agentic processes with near-zero error rates.

Use Cases:

  • Long sequences
  • Organization-level tasks
  • High-reliability requirements
  • Million-step processes

adaptive-agent.json

Difficulty: Advanced | Setup: 25 minutes | Icon: 🎯

Intelligent meta-agent that automatically selects the best agent type for each task with quality validation.

Use Cases:

  • Auto-optimization
  • Quality assurance
  • Agent selection
  • Adaptive workflows

Category: Specialized Agents

hierarchical-agent.json

Difficulty: Advanced | Setup: 20 minutes | Icon: 👑

Master-worker pattern agent that delegates specialized subtasks to worker agents.

Use Cases:

  • Complex reports
  • Multi-domain tasks
  • Team coordination
  • Specialized workflows

Example:

$agent = TemplateManager::instantiate('hierarchical-agent', [
    'api_key' => getenv('ANTHROPIC_API_KEY')
]);

// Register workers
$agent->registerWorker('researcher', $researchAgent);
$agent->registerWorker('writer', $writerAgent);

$result = $agent->run('Research AI trends and write a report');

coordinator-agent.json

Difficulty: Advanced | Setup: 25 minutes | Icon: 🎭

Multi-agent orchestration and load balancing for distributed systems.

Use Cases:

  • Agent networks
  • Load balancing
  • Distributed systems
  • Multi-agent coordination

dialog-agent.json

Difficulty: Intermediate | Setup: 15 minutes | Icon: 💬

Conversational AI agent with context tracking and memory for natural multi-turn conversations.

Use Cases:

  • Customer service
  • Virtual assistants
  • Chat interfaces
  • Interactive support

Example:

$agent = TemplateManager::instantiate('dialog-agent', [
    'api_key' => getenv('ANTHROPIC_API_KEY'),
    'context_window' => 5
]);

$result1 = $agent->run('Hello, my name is Alice');
$result2 = $agent->run('What is my name?'); // Agent remembers context

intent-classifier-agent.json

Difficulty: Intermediate | Setup: 15 minutes | Icon: 🎯

Intent recognition and entity extraction agent for command routing.

Use Cases:

  • Command routing
  • Intent detection
  • Entity extraction
  • NLU systems

monitoring-agent.json

Difficulty: Intermediate | Setup: 20 minutes | Icon: 📊

System monitoring and anomaly detection agent for real-time performance tracking.

Use Cases:

  • System monitoring
  • Performance tracking
  • Anomaly detection
  • Incident management

Category: RAG & Knowledge

rag-agent.json

Difficulty: Intermediate | Setup: 20 minutes | Icon: 📚

Retrieval-Augmented Generation agent for document retrieval and question answering.

Use Cases:

  • Document Q&A
  • Knowledge retrieval
  • Information extraction
  • Context-aware responses

Example:

$agent = TemplateManager::instantiate('rag-agent', [
    'api_key' => getenv('ANTHROPIC_API_KEY'),
    'top_k' => 5
]);

// Add documents to vector store
$result = $agent->run('What are the key features of PHP 8.3?');

memory-chatbot.json

Difficulty: Intermediate | Setup: 20 minutes | Icon: 🧠

Conversational agent with persistent memory for maintaining context across sessions.

Use Cases:

  • Personal assistants
  • Long-term conversations
  • User preference tracking
  • Session continuity

knowledge-manager.json

Difficulty: Advanced | Setup: 25 minutes | Icon: 🗄️

Knowledge management and retrieval agent for organizing and accessing shared information.

Use Cases:

  • Knowledge bases
  • Document management
  • Information retrieval
  • Shared memory systems

Category: Workflows

sequential-tasks-agent.json

Difficulty: Intermediate | Setup: 15 minutes | Icon: 🔗

Multi-step workflow execution agent that processes tasks in sequence with state management.

Use Cases:

  • Data pipelines
  • Workflow automation
  • Sequential processing
  • State machines

Example:

$agent = TemplateManager::instantiate('sequential-tasks-agent', [
    'api_key' => getenv('ANTHROPIC_API_KEY')
]);

$result = $agent->run('Process user registration: validate email, create account, send welcome email');

debate-system.json

Difficulty: Advanced | Setup: 25 minutes | Icon: 🗣️

Multi-agent debate and consensus building system for solution refinement.

Use Cases:

  • Decision making
  • Solution refinement
  • Consensus building
  • Multi-perspective analysis

Category: Production

production-agent.json

Difficulty: Advanced | Setup: 30 minutes | Icon: 🏭

Production-ready agent with comprehensive error handling, logging, monitoring, and retry logic.

Features:

  • Retry logic with exponential backoff
  • Comprehensive error handling
  • Structured logging
  • Performance monitoring
  • Health checks

Use Cases:

  • Production deployments
  • Enterprise applications
  • Mission-critical tasks
  • High-reliability systems

Example:

$agent = TemplateManager::instantiate('production-agent', [
    'api_key' => getenv('ANTHROPIC_API_KEY'),
    'retry_max_attempts' => 3,
    'retry_delay_ms' => 1000,
    'timeout' => 30.0
]);

$agent->onError(function($error, $attempt) {
    logger()->error('Agent error', [
        'error' => $error->getMessage(),
        'attempt' => $attempt
    ]);
});

$result = $agent->run('Critical business task');

async-batch-processor.json

Difficulty: Advanced | Setup: 25 minutes | Icon: ⚡

Concurrent task processing agent using AMPHP for high-throughput parallel execution.

Features:

  • Concurrent execution
  • Batch processing
  • Progress tracking
  • Error handling per task

Use Cases:

  • Batch processing
  • Concurrent execution
  • High-throughput tasks
  • Parallel workflows

Example:

$agent = TemplateManager::instantiate('async-batch-processor', [
    'api_key' => getenv('ANTHROPIC_API_KEY'),
    'concurrency' => 5,
    'batch_size' => 10
]);

$tasks = [
    'Process document 1',
    'Process document 2',
    // ... more tasks
];

$results = $agent->processBatch($tasks);

Template Comparison

By Difficulty

Beginner (2 templates)

  • Basic Agent
  • Reflex Agent

Intermediate (9 templates)

  • ReAct Agent
  • Chain-of-Thought Agent
  • Model-Based Agent
  • Dialog Agent
  • Intent Classifier Agent
  • Monitoring Agent
  • RAG Agent
  • Memory Chatbot
  • Sequential Tasks Agent

Advanced (11 templates)

  • Reflection Agent
  • Plan-Execute Agent
  • Tree-of-Thoughts Agent
  • MAKER Agent
  • Adaptive Agent
  • Hierarchical Agent
  • Coordinator Agent
  • Knowledge Manager
  • Debate System
  • Production Agent
  • Async Batch Processor

By Setup Time

Quick (≤10 minutes): 5 templates

  • Basic Agent
  • ReAct Agent
  • Chain-of-Thought Agent
  • Reflex Agent
  • Sequential Tasks Agent

Medium (15-20 minutes): 10 templates

  • Model-Based Agent
  • Reflection Agent
  • Plan-Execute Agent
  • Tree-of-Thoughts Agent
  • MAKER Agent
  • Hierarchical Agent
  • Dialog Agent
  • Intent Classifier Agent
  • Monitoring Agent
  • RAG Agent
  • Memory Chatbot

Extended (25-30 minutes): 7 templates

  • Adaptive Agent
  • Coordinator Agent
  • Knowledge Manager
  • Debate System
  • Async Batch Processor
  • Production Agent

By Scalability

Template Max Steps Scalability
Reflex Agent ~10 Instant
Basic Agent ~100 Low
ReAct Agent ~500 Medium
Plan-Execute Agent ~1,000 High
Hierarchical Agent ~5,000 Very High
MAKER Agent 1,000,000+ Extreme

See Also