synapsekit 1.4.0__tar.gz → 1.4.2__tar.gz

{synapsekit-1.4.0 → synapsekit-1.4.2}/.all-contributorsrc

@@ -49,6 +49,13 @@
       "avatar_url": "https://avatars.githubusercontent.com/u/136477030?v=4",
       "profile": "https://github.com/DhruvGarg111",
       "contributions": ["code"]
+    },
+    {
+      "login": "adaumsilva",
+      "name": "Adam Silva",
+      "avatar_url": "https://avatars.githubusercontent.com/u/178027480?v=4",
+      "profile": "https://github.com/adaumsilva",
+      "contributions": ["code"]
     }
   ],
   "contributorsPerLine": 7

{synapsekit-1.4.0 → synapsekit-1.4.2}/CHANGELOG.md

@@ -7,6 +7,41 @@ SynapseKit uses [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 ---
 
+## [1.4.2] — 2026-03-28
+
+### Added
+
+- **HuggingFaceLLM** — Hugging Face Inference API via `AsyncInferenceClient`; supports serverless API (model ID) and Dedicated Inference Endpoints (URL); streaming and generate
+- **DynamoDBCacheBackend** — serverless LLM response caching on AWS DynamoDB; configurable partition key, TTL via `ttl_seconds`, custom region and endpoint
+- **MemcachedCacheBackend** — high-throughput distributed LLM caching via `aiomcache`; TTL via `exptime`; async-native
+- **GoogleSearchTool** — Google web search via SerpAPI (`google-search-results`); graceful handling of missing keys, empty results, and API errors
+- **Graph versioning and checkpoint migration** — `StateGraph(version="1", migrations={...})`; `CompiledGraph.resume()` now applies migration chains when loading older checkpoint versions; supports direct and chained `(next_version, state_dict)` migrations; missing paths raise `GraphRuntimeError`
+- 11 new tests (1368 total)
+
+### Improved
+
+- **SQLQueryTool** — added `params` dict for parameterized queries (eliminates string interpolation); `max_rows` cap; fixed variable name shadowing
+
+---
+
+## [1.4.1] — 2026-03-27
+
+### Added
+
+- **MinimaxLLM provider** — `MinimaxLLM` for Minimax API with SSE streaming; requires `group_id`; auto-detected from `minimax-*` model prefix
+- **AlephAlphaLLM provider** — `AlephAlphaLLM` for Aleph Alpha Luminous and Pharia models; httpx-based streaming; auto-detected from `luminous-*`/`pharia-*` prefixes
+- **YAMLLoader** — load YAML files (list-of-objects or single-object) into `Document` list; configurable `text_key` and `metadata_keys`; uses `yaml.safe_load()`
+- **BingSearchTool** — web search via Bing Web Search API v7; auth via `Ocp-Apim-Subscription-Key`; supports `query`, `count` (capped at 50), `market`
+- **WolframAlphaTool** — computational queries via Wolfram Alpha API; short-answer endpoint; thread-executor async wrapper
+- **Usage examples** — `examples/` directory with 5 runnable scripts: RAG quickstart, agent with tools, graph workflow, multi-provider, caching & retries
+- 30 new tests (1357 total)
+
+### Fixed
+
+- Added missing return type annotations to `_loader_for()` in `loaders/directory.py` and `__getattr__()` in `loaders/__init__.py`
+
+---
+
 ## [1.4.0] — 2026-03-25
 
 ### Added

{synapsekit-1.4.0 → synapsekit-1.4.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: synapsekit
-Version: 1.4.0
+Version: 1.4.2
 Summary: Async-native Python framework for building production-grade LLM applications. Streaming-first, 2 dependencies, fully transparent.
 Project-URL: Homepage, https://github.com/SynapseKit/SynapseKit
 Project-URL: Repository, https://github.com/SynapseKit/SynapseKit
@@ -23,8 +23,11 @@ Requires-Dist: numpy>=1.26
 Requires-Dist: rank-bm25>=0.2.2
 Provides-Extra: ai21
 Requires-Dist: ai21>=2.0; extra == 'ai21'
+Provides-Extra: aleph-alpha
+Requires-Dist: aleph-alpha-client>=7.0; extra == 'aleph-alpha'
 Provides-Extra: all
 Requires-Dist: ai21>=2.0; extra == 'all'
+Requires-Dist: aiomcache>=0.8; extra == 'all'
 Requires-Dist: anthropic>=0.25; extra == 'all'
 Requires-Dist: beautifulsoup4>=4.12; extra == 'all'
 Requires-Dist: boto3>=1.34; extra == 'all'
@@ -36,8 +39,10 @@ Requires-Dist: faiss-cpu>=1.7; extra == 'all'
 Requires-Dist: fastapi>=0.110; extra == 'all'
 Requires-Dist: google-cloud-aiplatform>=1.38; extra == 'all'
 Requires-Dist: google-generativeai>=0.7; extra == 'all'
+Requires-Dist: google-search-results>=2.4; extra == 'all'
 Requires-Dist: groq>=0.9; extra == 'all'
 Requires-Dist: httpx>=0.27; extra == 'all'
+Requires-Dist: huggingface-hub>=0.20; extra == 'all'
 Requires-Dist: llama-cpp-python>=0.2; extra == 'all'
 Requires-Dist: lxml>=5.0; extra == 'all'
 Requires-Dist: mcp>=1.0; extra == 'all'
@@ -47,11 +52,13 @@ Requires-Dist: openai>=1.0; extra == 'all'
 Requires-Dist: pinecone>=3.0; extra == 'all'
 Requires-Dist: psycopg[binary]>=3.1; extra == 'all'
 Requires-Dist: pypdf>=4.0; extra == 'all'
+Requires-Dist: pyyaml>=6.0; extra == 'all'
 Requires-Dist: qdrant-client>=1.9; extra == 'all'
 Requires-Dist: redis>=5.0; extra == 'all'
 Requires-Dist: sentence-transformers>=2.7; extra == 'all'
 Requires-Dist: tavily-python>=0.3; extra == 'all'
 Requires-Dist: uvicorn[standard]>=0.29; extra == 'all'
+Requires-Dist: wolframalpha>=5.0; extra == 'all'
 Requires-Dist: youtube-search-python>=1.6; extra == 'all'
 Provides-Extra: anthropic
 Requires-Dist: anthropic>=0.25; extra == 'anthropic'
@@ -69,6 +76,8 @@ Provides-Extra: cohere
 Requires-Dist: cohere>=5.0; extra == 'cohere'
 Provides-Extra: docx
 Requires-Dist: python-docx>=1.0; extra == 'docx'
+Provides-Extra: dynamodb
+Requires-Dist: boto3>=1.34; extra == 'dynamodb'
 Provides-Extra: ernie
 Requires-Dist: erniebot>=0.5; extra == 'ernie'
 Provides-Extra: excel
@@ -79,6 +88,8 @@ Provides-Extra: gcal-tool
 Requires-Dist: google-api-python-client>=2.0; extra == 'gcal-tool'
 Provides-Extra: gemini
 Requires-Dist: google-generativeai>=0.7; extra == 'gemini'
+Provides-Extra: google-search
+Requires-Dist: google-search-results>=2.4; extra == 'google-search'
 Provides-Extra: groq
 Requires-Dist: groq>=0.9; extra == 'groq'
 Provides-Extra: html
@@ -86,10 +97,16 @@ Requires-Dist: beautifulsoup4>=4.12; extra == 'html'
 Requires-Dist: lxml>=5.0; extra == 'html'
 Provides-Extra: http
 Requires-Dist: aiohttp>=3.9; extra == 'http'
+Provides-Extra: huggingface
+Requires-Dist: huggingface-hub>=0.20; extra == 'huggingface'
 Provides-Extra: llamacpp
 Requires-Dist: llama-cpp-python>=0.2; extra == 'llamacpp'
 Provides-Extra: mcp
 Requires-Dist: mcp>=1.0; extra == 'mcp'
+Provides-Extra: memcached
+Requires-Dist: aiomcache>=0.8; extra == 'memcached'
+Provides-Extra: minimax
+Requires-Dist: httpx>=0.27; extra == 'minimax'
 Provides-Extra: mistral
 Requires-Dist: mistralai>=1.0; extra == 'mistral'
 Provides-Extra: ollama
@@ -124,6 +141,10 @@ Requires-Dist: openai>=1.0; extra == 'video'
 Provides-Extra: web
 Requires-Dist: beautifulsoup4>=4.12; extra == 'web'
 Requires-Dist: httpx>=0.27; extra == 'web'
+Provides-Extra: wolfram
+Requires-Dist: wolframalpha>=5.0; extra == 'wolfram'
+Provides-Extra: yaml
+Requires-Dist: pyyaml>=6.0; extra == 'yaml'
 Provides-Extra: youtube
 Requires-Dist: youtube-search-python>=1.6; extra == 'youtube'
 Description-Content-Type: text/markdown
@@ -334,6 +355,7 @@ Read [CONTRIBUTING.md](CONTRIBUTING.md) to get started. Look for issues tagged [
       <td align="center" valign="top" width="14.28%"><a href="https://github.com/Arjunkundapur"><img src="https://avatars.githubusercontent.com/u/64265396?v=4" width="100px;" alt="Arjun Kundapur"/><br /><sub><b>Arjun Kundapur</b></sub></a><br /><a href="https://github.com/SynapseKit/SynapseKit/commits?author=Arjunkundapur" title="Code">💻</a></td>
       <td align="center" valign="top" width="14.28%"><a href="https://github.com/Ashusf90"><img src="https://avatars.githubusercontent.com/u/153393197?v=4" width="100px;" alt="Harshit Gupta"/><br /><sub><b>Harshit Gupta</b></sub></a><br /><a href="https://github.com/SynapseKit/synapsekit-docs/pull/34" title="Documentation">📖</a></td>
       <td align="center" valign="top" width="14.28%"><a href="https://github.com/DhruvGarg111"><img src="https://avatars.githubusercontent.com/u/136477030?v=4" width="100px;" alt="Dhruv Garg"/><br /><sub><b>Dhruv Garg</b></sub></a><br /><a href="https://github.com/SynapseKit/SynapseKit/commits?author=DhruvGarg111" title="Code">💻</a></td>
+      <td align="center" valign="top" width="14.28%"><a href="https://github.com/adaumsilva"><img src="https://avatars.githubusercontent.com/u/178027480?v=4" width="100px;" alt="Adam Silva"/><br /><sub><b>Adam Silva</b></sub></a><br /><a href="https://github.com/SynapseKit/SynapseKit/commits?author=adaumsilva" title="Code">💻</a></td>
     </tr>
   </tbody>
 </table>

{synapsekit-1.4.0 → synapsekit-1.4.2}/README.md

@@ -204,6 +204,7 @@ Read [CONTRIBUTING.md](CONTRIBUTING.md) to get started. Look for issues tagged [
       <td align="center" valign="top" width="14.28%"><a href="https://github.com/Arjunkundapur"><img src="https://avatars.githubusercontent.com/u/64265396?v=4" width="100px;" alt="Arjun Kundapur"/><br /><sub><b>Arjun Kundapur</b></sub></a><br /><a href="https://github.com/SynapseKit/SynapseKit/commits?author=Arjunkundapur" title="Code">💻</a></td>
       <td align="center" valign="top" width="14.28%"><a href="https://github.com/Ashusf90"><img src="https://avatars.githubusercontent.com/u/153393197?v=4" width="100px;" alt="Harshit Gupta"/><br /><sub><b>Harshit Gupta</b></sub></a><br /><a href="https://github.com/SynapseKit/synapsekit-docs/pull/34" title="Documentation">📖</a></td>
       <td align="center" valign="top" width="14.28%"><a href="https://github.com/DhruvGarg111"><img src="https://avatars.githubusercontent.com/u/136477030?v=4" width="100px;" alt="Dhruv Garg"/><br /><sub><b>Dhruv Garg</b></sub></a><br /><a href="https://github.com/SynapseKit/SynapseKit/commits?author=DhruvGarg111" title="Code">💻</a></td>
+      <td align="center" valign="top" width="14.28%"><a href="https://github.com/adaumsilva"><img src="https://avatars.githubusercontent.com/u/178027480?v=4" width="100px;" alt="Adam Silva"/><br /><sub><b>Adam Silva</b></sub></a><br /><a href="https://github.com/SynapseKit/SynapseKit/commits?author=adaumsilva" title="Code">💻</a></td>
     </tr>
   </tbody>
 </table>

synapsekit-1.4.2/examples/README.md

@@ -0,0 +1,70 @@
+# Examples
+
+This directory contains runnable examples demonstrating key SynapseKit features.
+
+## Prerequisites
+
+Install SynapseKit with OpenAI support:
+```bash
+pip install synapsekit[openai]
+```
+
+Set your API key:
+```bash
+export OPENAI_API_KEY=sk-...
+```
+
+## Examples
+
+### 1. `rag_quickstart.py` — RAG Basics
+The simplest way to get started: load text, add documents, and query with streaming.
+
+```bash
+python examples/rag_quickstart.py
+```
+
+### 2. `agent_tools.py` — ReAct Agent with Tools
+Create a ReAct agent with built-in and custom tools. Shows reasoning and tool execution.
+
+```bash
+python examples/agent_tools.py
+```
+
+### 3. `graph_workflow.py` — State Graph with Conditional Routing
+Build workflows with state management, conditional edges, and visualization.
+
+```bash
+python examples/graph_workflow.py
+```
+
+### 4. `multi_provider.py` — Multi-Provider Comparison
+Run the same prompt across OpenAI, Anthropic, and Ollama to compare responses.
+
+Requires additional setup:
+```bash
+pip install synapsekit[openai,anthropic]
+export ANTHROPIC_API_KEY=sk-ant-...
+```
+
+```bash
+python examples/multi_provider.py
+```
+
+### 5. `caching_retries.py` — Advanced LLM Configuration
+Configure response caching, automatic retries, and cost tracking with budget limits.
+
+```bash
+python examples/caching_retries.py
+```
+
+## General Pattern
+
+All examples follow this pattern:
+- Use `os.environ` for API keys (never hardcode)
+- Include docstrings explaining what the example does
+- Work with minimal dependencies (`pip install synapsekit[openai]`)
+- Print step-by-step progress for learning
+
+## Contributing
+
+Found an issue or want to add more examples? Open an issue or PR on GitHub!

synapsekit-1.4.2/examples/agent_tools.py

@@ -0,0 +1,95 @@
+"""
+ReAct Agent with Custom Tools
+==============================
+
+This example shows how to create a ReAct agent with built-in and custom tools.
+The agent can reason about which tool to use and execute them step-by-step.
+
+Prerequisites:
+    pip install synapsekit[openai]
+
+Usage:
+    export OPENAI_API_KEY=sk-...
+    python examples/agent_tools.py
+"""
+
+import asyncio
+import os
+from datetime import datetime
+
+from synapsekit import ReActAgent, tool, CalculatorTool, DateTimeTool
+
+
+# Define a custom tool
+@tool
+def get_weather(location: str) -> str:
+    """
+    Get current weather for a location.
+    
+    Args:
+        location: City name or location
+        
+    Returns:
+        Weather description
+    """
+    # Mock weather data
+    return f"The weather in {location} is sunny with 22°C temperature."
+
+
+@tool
+def convert_currency(amount: float, from_currency: str, to_currency: str) -> str:
+    """
+    Convert currency from one type to another.
+    
+    Args:
+        amount: Amount to convert
+        from_currency: Source currency code (e.g., USD, EUR)
+        to_currency: Target currency code
+        
+    Returns:
+        Converted amount as string
+    """
+    # Mock conversion (in reality, fetch real rates)
+    rates = {"USD": 1.0, "EUR": 0.85, "GBP": 0.73, "JPY": 110.0}
+    if from_currency not in rates or to_currency not in rates:
+        return "Currency not supported"
+    
+    result = amount * (rates[to_currency] / rates[from_currency])
+    return f"{amount} {from_currency} = {result:.2f} {to_currency}"
+
+
+async def main():
+    api_key = os.environ.get("OPENAI_API_KEY")
+    if not api_key:
+        raise ValueError("Please set OPENAI_API_KEY environment variable")
+
+    # Create agent with multiple tools
+    agent = ReActAgent(
+        model="gpt-4o-mini",
+        api_key=api_key,
+        tools=[
+            CalculatorTool(),
+            DateTimeTool(),
+            get_weather,
+            convert_currency,
+        ],
+        verbose=True,
+    )
+
+    # Example queries that require different tools
+    queries = [
+        "What's the weather in Tokyo and how much is 100 USD in JPY?",
+        "Calculate 15% tip on a $45.50 bill and tell me today's date.",
+    ]
+
+    for query in queries:
+        print(f"\n{'='*60}")
+        print(f"Query: {query}")
+        print('='*60)
+        
+        result = await agent.run(query)
+        print(f"\nFinal Answer: {result}")
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

synapsekit-1.4.2/examples/caching_retries.py

@@ -0,0 +1,171 @@
+"""
+LLM with Caching and Retry Configuration
+=========================================
+
+This example demonstrates how to configure LLM calls with:
+1. Response caching for repeated queries
+2. Automatic retries with exponential backoff
+3. Cost tracking and budget limits
+
+Prerequisites:
+    pip install synapsekit[openai]
+
+Usage:
+    export OPENAI_API_KEY=sk-...
+    python examples/caching_retries.py
+"""
+
+import asyncio
+import os
+from datetime import datetime
+
+from synapsekit import BaseLLM, LLMConfig, CostTracker, BudgetGuard
+
+
+# Simple in-memory cache implementation
+class SimpleCache:
+    """Simple in-memory cache for demonstration"""
+    
+    def __init__(self):
+        self.cache = {}
+    
+    def get(self, key: str):
+        """Get cached value"""
+        return self.cache.get(key)
+    
+    def set(self, key: str, value: str):
+        """Set cached value"""
+        self.cache[key] = value
+        print(f"💾 Cached response for: {key[:50]}...")
+    
+    def clear(self):
+        """Clear cache"""
+        self.cache.clear()
+        print("🗑️  Cache cleared")
+
+
+async def demo_caching():
+    """Demonstrate response caching"""
+    print("\n" + "="*60)
+    print("Demo 1: Response Caching")
+    print("="*60 + "\n")
+    
+    api_key = os.environ.get("OPENAI_API_KEY")
+    if not api_key:
+        print("⚠️  OPENAI_API_KEY not set. Skipping demo.")
+        return
+    
+    from synapsekit.llm.openai import OpenAILLM
+    
+    llm = OpenAILLM(model="gpt-4o-mini", api_key=api_key)
+    cache = SimpleCache()
+    
+    prompt = "What is the capital of France?"
+    
+    # First call (no cache)
+    print(f"Query: {prompt}")
+    start = datetime.now()
+    
+    cached = cache.get(prompt)
+    if cached:
+        print("✅ Using cached response")
+        response = cached
+    else:
+        print("🔄 Making API call...")
+        response = await llm.agenerate(prompt)
+        cache.set(prompt, response)
+    
+    elapsed = (datetime.now() - start).total_seconds()
+    print(f"Response: {response}")
+    print(f"⏱️  Time: {elapsed:.2f}s\n")
+    
+    # Second call (cached)
+    print(f"Query (repeated): {prompt}")
+    start = datetime.now()
+    
+    cached = cache.get(prompt)
+    if cached:
+        print("✅ Using cached response")
+        response = cached
+    else:
+        print("🔄 Making API call...")
+        response = await llm.agenerate(prompt)
+        cache.set(prompt, response)
+    
+    elapsed = (datetime.now() - start).total_seconds()
+    print(f"Response: {response}")
+    print(f"⏱️  Time: {elapsed:.2f}s (much faster!)\n")
+
+
+async def demo_retries():
+    """Demonstrate retry configuration"""
+    print("\n" + "="*60)
+    print("Demo 2: Retry Configuration")
+    print("="*60 + "\n")
+    
+    print("🔧 LLM Config with retries:")
+    print("  - Max retries: 3")
+    print("  - Timeout: 30s")
+    print("  - Exponential backoff\n")
+    
+    # In a real scenario, configure via LLMConfig
+    config = LLMConfig(
+        max_retries=3,
+        timeout=30,
+        retry_on_errors=["RateLimitError", "TimeoutError"],
+    )
+    
+    print(f"Config: {config}\n")
+    print("💡 On API failures, requests will retry automatically with backoff.")
+
+
+async def demo_cost_tracking():
+    """Demonstrate cost tracking and budgets"""
+    print("\n" + "="*60)
+    print("Demo 3: Cost Tracking & Budget Limits")
+    print("="*60 + "\n")
+    
+    # Initialize cost tracker
+    tracker = CostTracker()
+    
+    # Simulate some API calls
+    tracker.add_tokens(model="gpt-4o-mini", prompt_tokens=100, completion_tokens=50)
+    tracker.add_tokens(model="gpt-4o-mini", prompt_tokens=200, completion_tokens=75)
+    tracker.add_tokens(model="gpt-4o-mini", prompt_tokens=150, completion_tokens=60)
+    
+    print("📊 Cost Summary:")
+    print(f"  Total tokens: {tracker.total_tokens}")
+    print(f"  Total cost: ${tracker.total_cost:.4f}\n")
+    
+    # Set up budget guard
+    print("🛡️  Setting budget limit: $5.00")
+    budget = BudgetGuard(limit=5.0)
+    
+    try:
+        # Check if under budget
+        current_cost = tracker.total_cost
+        if budget.check(current_cost):
+            print(f"✅ Under budget (${current_cost:.4f} / $5.00)")
+        else:
+            print(f"❌ Budget exceeded!")
+    except Exception as e:
+        print(f"⚠️  {e}")
+
+
+async def main():
+    print("""
+This example demonstrates three key LLM configuration patterns:
+1. Response caching for repeated queries
+2. Automatic retry with exponential backoff
+3. Cost tracking and budget limits
+    """)
+    
+    await demo_caching()
+    await demo_retries()
+    await demo_cost_tracking()
+    
+    print("\n✨ All demos complete!\n")
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

synapsekit-1.4.2/examples/graph_workflow.py

@@ -0,0 +1,125 @@
+"""
+Graph Workflow with Conditional Edges
+======================================
+
+This example demonstrates SynapseKit's StateGraph for building workflows
+with conditional routing, state management, and visualization.
+
+Prerequisites:
+    pip install synapsekit[openai]
+
+Usage:
+    export OPENAI_API_KEY=sk-...
+    python examples/graph_workflow.py
+"""
+
+import asyncio
+import os
+from typing import Literal
+
+from synapsekit import StateGraph, END
+
+
+# Define workflow state
+class ReviewState:
+    """State for content review workflow"""
+    content: str = ""
+    word_count: int = 0
+    is_approved: bool = False
+    feedback: str = ""
+
+
+# Define node functions
+async def count_words(state: ReviewState) -> ReviewState:
+    """Count words in content"""
+    words = state.content.split()
+    state.word_count = len(words)
+    print(f"📊 Word count: {state.word_count}")
+    return state
+
+
+async def check_length(state: ReviewState) -> ReviewState:
+    """Check if content meets length requirements"""
+    if state.word_count >= 10:
+        state.is_approved = True
+        state.feedback = "Content length is acceptable."
+    else:
+        state.is_approved = False
+        state.feedback = f"Content too short. Need {10 - state.word_count} more words."
+    
+    print(f"✅ Approved: {state.is_approved}")
+    print(f"💬 Feedback: {state.feedback}")
+    return state
+
+
+async def approve_content(state: ReviewState) -> ReviewState:
+    """Approve content for publication"""
+    print("✨ Content approved for publication!")
+    return state
+
+
+async def reject_content(state: ReviewState) -> ReviewState:
+    """Reject content and request revision"""
+    print("❌ Content rejected. Revision needed.")
+    return state
+
+
+# Conditional routing function
+def should_approve(state: ReviewState) -> Literal["approve", "reject"]:
+    """Route based on approval status"""
+    return "approve" if state.is_approved else "reject"
+
+
+async def main():
+    # Build the workflow graph
+    graph = StateGraph(ReviewState)
+
+    # Add nodes
+    graph.add_node("count", count_words)
+    graph.add_node("check", check_length)
+    graph.add_node("approve", approve_content)
+    graph.add_node("reject", reject_content)
+
+    # Add edges
+    graph.set_entry_point("count")
+    graph.add_edge("count", "check")
+    
+    # Add conditional edge based on approval
+    graph.add_conditional_edges(
+        "check",
+        should_approve,
+        {
+            "approve": "approve",
+            "reject": "reject",
+        }
+    )
+    
+    # Both approval and rejection lead to end
+    graph.add_edge("approve", END)
+    graph.add_edge("reject", END)
+
+    # Compile the graph
+    workflow = graph.compile()
+
+    # Visualize the graph (optional)
+    print("Graph structure:")
+    print(workflow.get_mermaid())
+    print("\n" + "="*60 + "\n")
+
+    # Test with different inputs
+    test_contents = [
+        "Hello world!",  # Too short
+        "This is a longer piece of content that meets the minimum word count requirement.",  # Long enough
+    ]
+
+    for i, content in enumerate(test_contents, 1):
+        print(f"Test {i}: {content[:50]}...")
+        print("-" * 60)
+        
+        result = await workflow.run(ReviewState(content=content))
+        print(f"\nFinal state - Approved: {result.is_approved}")
+        print("="*60 + "\n")
+
+
+if __name__ == "__main__":
+    asyncio.run(main())