Two agents in a row

In a sequential crew, tasks run in list order and each one sees the output of those before it. The clerk finds the order; the writer answers from that.

Lesson 2's desk had two workers: a clerk and a writer who only saw the clerk's finding. A crew with two agents and two tasks does the same.

from crewai import Agent, Crew, Task
from shop_llm import ShopLLM
from tools import lookup_order

clerk = Agent(role="Order clerk", goal="Find the status of customers' orders",
              backstory="You can look up any order in the shop's system.",
              llm=ShopLLM(model="shop"), tools=[lookup_order])
writer = Agent(role="Reply writer", goal="Write replies to customers",
               backstory="You write short, friendly emails.",
               llm=ShopLLM(model="shop"))

The clerk has the lookup tool. The writer has none: its job is words.

look = Task(description="Find the order in this message: {question}",
            expected_output="The order's status.", agent=clerk)
reply = Task(description="Write the customer a reply.",
             expected_output="A short, friendly email.", agent=writer)
crew = Crew(agents=[clerk, writer], tasks=[look, reply])

The reply task does not mention the question. It gets the clerk's answer instead, because a crew's default process, sequential, runs tasks in list order and passes the earlier outputs to each later task.

result = crew.kickoff(inputs={"question": "Where is my order A17?"})
print(result.raw)

The writer's model received the clerk's output under the heading This is the context you're working with:, and lesson 9's model has no rule for that, so it saw A17, no tool, and gave up.

A model that reads the context

        if "working with:" in text:
            context = text.split("working with:")[1].strip().split("\n\n")[0]
            return f"Dear customer, {context}"

When a message carries context and no tool fits, the model writes a reply from it. Put these lines in decide after the tool call and before if orders:, and save the file; the lessons from here on use this version.

result = crew.kickoff(inputs={"question": "Where is my order A17?"})

for output in result.tasks_output:
    print(f"{output.agent}: {output.raw}")

tasks_output shows both steps: the clerk's finding, then the writer's email built from it. result.raw is the last task's output.

Choosing what a task sees

reply.context = [look]
result = crew.kickoff(inputs={"question": "Where is my order B22?"})
print(result.raw)

context names the tasks whose outputs a task receives. Here it matches the default, but with three tasks it lets the last one see the first and skip the middle. The tasks page asks for it whenever a task depends on one that did not run immediately before it.