feat: python agent toolkit wrapper + langchain integration

.github/workflows/publish.yml

@@ -23,6 +23,8 @@ jobs:
       - name: Check package version and detect an update
         id: check-package-version
         uses: PostHog/check-package-version@v2
+        with:
+          path: typescript/
 
   release:
     name: Publish release if new version

.gitignore

@@ -17,4 +17,6 @@ __pycache__/
 
 .env
 
-dist/
+dist/
+
+.venv/

examples/langchain/.env.example

@@ -0,0 +1,6 @@
+# PostHog Configuration
+POSTHOG_PERSONAL_API_KEY=your_posthog_api_key_here
+POSTHOG_MCP_URL=https://mcp.posthog.com/mcp  # Optional
+
+# OpenAI Configuration (for LangChain agent)
+OPENAI_API_KEY=your_openai_api_key_here

examples/langchain/README.md

@@ -0,0 +1,36 @@
+# PostHog LangChain Python Integration Example
+
+This example demonstrates how to use PostHog tools with LangChain using the `posthog_agent_toolkit` package.
+
+## Setup
+
+1. Install dependencies:
+```bash
+# Reinstall the local PostHog Agent Toolkit to ensure latest changes
+uv sync --reinstall-package posthog-agent-toolkit
+```
+
+2. Copy the environment file and fill in your credentials:
+```bash
+cp .env.example .env
+```
+
+3. Update your `.env` file with:
+   - `POSTHOG_PERSONAL_API_KEY`: Your PostHog personal API key
+   - `OPENAI_API_KEY`: Your OpenAI API key
+
+## Usage
+
+Run the example:
+```bash
+uv run python posthog_agent_example.py
+```
+
+The example will:
+1. Connect to the PostHog MCP server using your personal API key
+2. Load all available PostHog tools from the MCP server
+3. Create a LangChain agent
+4. Analyze product usage by:
+   - Getting available insights
+   - Querying data for the most relevant ones
+   - Summarizing key findings

examples/langchain/posthog_agent_example.py

@@ -0,0 +1,103 @@
+"""
+PostHog LangChain Integration Example
+
+This example demonstrates how to use PostHog tools with LangChain using
+the local posthog_agent_toolkit package. It shows how to analyze product
+usage data similar to the TypeScript example.
+"""
+
+import asyncio
+import os
+import sys
+
+from dotenv import load_dotenv
+from langchain_openai import ChatOpenAI
+from langchain.agents import AgentExecutor, create_tool_calling_agent
+from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
+from posthog_agent_toolkit.integrations.langchain.toolkit import PostHogAgentToolkit
+
+
+async def analyze_product_usage():
+    """Analyze product usage using PostHog data."""
+
+    print("🚀 PostHog LangChain Agent - Product Usage Analysis\n")
+
+    # Initialize the PostHog toolkit with credentials
+    toolkit = PostHogAgentToolkit(
+        personal_api_key=os.getenv("POSTHOG_PERSONAL_API_KEY"),
+        url=os.getenv("POSTHOG_MCP_URL", "https://mcp.posthog.com/mcp")
+    )
+
+    # Get the tools
+    tools = await toolkit.get_tools()
+
+    # Initialize the LLM
+    llm = ChatOpenAI(
+        model="gpt-4o-mini",
+        temperature=0,
+        api_key=os.getenv("OPENAI_API_KEY")
+    )
+
+    # Create a system prompt for the agent
+    prompt = ChatPromptTemplate.from_messages([
+        (
+            "system",
+            "You are a data analyst. Your task is to do a deep dive into what's happening in our product. "
+            "Be concise and data-driven in your responses."
+        ),
+        ("human", "{input}"),
+        MessagesPlaceholder("agent_scratchpad"),
+    ])
+
+    agent = create_tool_calling_agent(
+        llm=llm,
+        tools=tools,
+        prompt=prompt,
+    )
+
+    agent_executor = AgentExecutor(
+        agent=agent,
+        tools=tools,
+        verbose=False,
+        max_iterations=5,
+    )
+
+    # Invoke the agent with an analysis request
+    result = await agent_executor.ainvoke({
+        "input": """Please analyze our product usage:
+
+        1. Get all available insights (limit 100)
+        2. Pick the 5 MOST INTERESTING and VALUABLE insights - prioritize:
+           - User behavior and engagement metrics
+           - Conversion funnels
+           - Retention and growth metrics
+           - Product adoption insights
+           - Revenue or business KPIs
+           AVOID picking feature flag insights unless they show significant business impact
+        3. For each selected insight, query its data and explain why it's important
+        4. Summarize the key findings in a brief report with actionable recommendations
+
+        Focus on insights that tell a story about user behavior and business performance."""
+    })
+
+    print("\n📊 Analysis Complete!\n")
+    print("=" * 50)
+    print(result["output"])
+    print("=" * 50)
+
+
+async def main():
+    """Main function to run the product usage analysis."""
+    try:
+        # Load environment variables
+        load_dotenv()
+
+        # Run the analysis
+        await analyze_product_usage()
+    except Exception as error:
+        print(f"Error: {error}")
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

examples/langchain/pyproject.toml

@@ -0,0 +1,24 @@
+[project]
+name = "posthog-langchain-example"
+version = "0.1.0"
+description = "PostHog LangChain integration example"
+readme = "README.md"
+requires-python = ">=3.11"
+dependencies = [
+    "posthog-agent-toolkit",
+    "langchain>=0.3.0",
+    "langchain-openai>=0.2.0",
+    "langchain-core>=0.3.0",
+    "python-dotenv>=1.0.0",
+]
+
+[tool.uv.sources]
+posthog-agent-toolkit = { path = "../../python", editable = true }
+
+[tool.ruff]
+line-length = 120
+target-version = "py311"
+
+[tool.black]
+line-length = 120
+target-version = ['py311']