Langchain v0.3 React Agent Error: Missing Required Prompt Variables

The error

  ValueError: Prompt missing required variables: {'tool_names', 'agent_scratchpad', 'tools'}

indicates that the prompt you supplied doesn’t include the required placeholders for the React agent in LangChain v0.3. In this version, the React agent expects your prompt to include variables that it uses to manage tool invocation and internal reasoning. Specifically, it requires placeholders for:

• tool_names
• agent_scratchpad
• tools

In your code, your prompt only has {tools_description} and {input}. To resolve the issue, you need to update your prompt template to include the required variables.

Below is an updated example of how you might modify your prompt:

---

PYTHONCopied
from langchain_openai import ChatOpenAI
from langchain.agents.react.agent import create_react_agent
from langchain_community.tools import Tool
from langchain.prompts import PromptTemplate

# Define a mock function to find similar companies
def find_similar_companies(company_name: str) -> str:
    mock_data = {
        "XYZ Corp": "ABC Inc, DEF Ltd",
        "Acme Corp": "GHI LLC, JKL PLC"
    }
    return f"Similar companies to {company_name}: {mock_data.get(company_name, 'No similar companies found.')}"

# Define the tools using the Tool class
tools = [
    Tool(
        name="Find Similar Companies",
        func=find_similar_companies,
        description="Use this tool to find similar companies based on a company name."
    )
]

# Initialize the ChatOpenAI model
llm = ChatOpenAI(
    model="gpt-3.5-turbo",
    openai_api_key="API_KEY"
)

# Create a PromptTemplate including the required variables: input, tool_names, agent_scratchpad, and tools.
# Here, you can dynamically fill in the tool names and tool descriptions if needed.
tool_names = ", ".join([t.name for t in tools])
tools_description = "; ".join([t.description for t in tools])

prompt_template = PromptTemplate(
    template="""
You are an intelligent agent. Use the available tools when needed.
Tool Names: {tool_names}
Tools: {tools}
Remember to document any internal thoughts in your scratchpad: {agent_scratchpad}
Query: {input}
""",
    input_variables=["input", "tool_names", "agent_scratchpad", "tools"]
)

# Create a React agent with the updated prompt template
agent = create_react_agent(
    llm=llm,
    tools=tools,
    prompt=prompt_template
)

# Test the agent
query = "Find similar companies to XYZ Corp"
# Supply initial values for the required variables.
# Note: The agent will manage the "agent_scratchpad" internally during the conversation.
response = agent.invoke({
    "input": query,
    "tool_names": tool_names,
    "tools": tools_description,
    "agent_scratchpad": ""
})

# Print the response
print("Agent Response:", response)

---

Key Points:

• Required Variables:
  Your prompt template must include placeholders for "tool_names", "agent_scratchpad", and "tools" because the React agent uses these to manage tool invocation and chain-of-thought reasoning.

• Dynamic Values:
  In the example above, we create dynamic strings for tool names and tool descriptions. During a real conversation, the agent will update the "agent_scratchpad" as it processes the query.

• State Management:
  LangChain’s React agent expects to inject its internal state into the prompt, hence providing an empty string for "agent_scratchpad" initially is common, as it will be updated with intermediate steps.

Adjust your prompt and initial call accordingly to resolve the error. This should get your simple agent working correctly in LangChain v0.3.