VAP LangChain Tools

Execution control for LangChain agents calling paid APIs.

LangChain tools for VAP - the execution control layer. Enforce cost limits and deterministic retries when your chains generate video, music, and images.

Installation

pip install vap-langchain

Quick Start

from langchain.agents import initialize_agent, AgentType
from langchain_openai import ChatOpenAI
from vap_langchain import VapProductionTool

# Initialize tools
vap_tool = VapProductionTool(api_key="vap_your_api_key")

# Create agent
llm = ChatOpenAI(model="gpt-4")
agent = initialize_agent(
    tools=[vap_tool],
    llm=llm,
    agent=AgentType.OPENAI_FUNCTIONS,
)

# Run
result = agent.run("Create an energetic startup launch video with upbeat music")
print(result)

Available Tools

Recommended: VapProductionTool

For most use cases, use VapProductionTool. It creates everything in one call:

from vap_langchain import VapProductionTool

tool = VapProductionTool(api_key="vap_...")

# Agent will automatically use this for any media request
result = tool.run("Cozy coffee shop morning scene with gentle acoustic vibes")
# Returns: Video URL, Music URL, Thumbnail URL

With LangGraph

from langgraph.graph import StateGraph
from vap_langchain import VapProductionTool

vap = VapProductionTool(api_key="vap_...")

def produce_media(state):
    result = vap.run(state["prompt"])
    return {"media": result}

# Add to your graph
graph.add_node("producer", produce_media)

Budget Safety

VAP uses a prepaid model - your agent can't overspend. The tool will return an error if balance is insufficient, preventing runaway costs.

Links

• VAP Documentation
• VAP SDK
• GitHub