Agent Not Using Observation's from one Tool for another Tool

[KyleGrande]

So I have a few PromptTemplates tied to an LLMChain as tools for my agent, however, as the agent uses the tools it only ever uses the context of the agent.run('a_context")

create_b_context = PromptTemplate(
    input_variables=["user_input"],
    template="Create some b context using {user_input}"
    )
create_c_context = PromptTemplate(
    input_variables=["user_input"],
    template="Create some b context using {user_input}"
    )

llm = OpenAI(openai_api_key=api_key)
llm = OpenAI(temperature=0.9)

chain_create_b_context= LLMChain(llm=llm, prompt=create_c_context)
chain_create_c_context = LLMChain(llm=llm, prompt=create_c_context)

tools = [
    Tool(
        name = 'create_b_context_tool',
        func=chain_create_b_context.run,
        description="Create B Context",
        ),
    Tool(
        name = 'create_c_context_tool',
        func=chain_create_c_context.run,
        description="Create C context using B Context",
        ),
]

agent = initialize_agent(tools, llm, agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION, verbose=True)

agent.run("A context")

The output is usually something like:

I need B and C Context
Action: create_b_context
Action Input: A Context
Observation: Context B
Thought: I have B, now I need C
Action: create_c_context
Action Input: A Context