Drill · lap 1 · 1/9 · AGL-101 · Pull the answer out of a model message

[akash-coded]

1 of 9 · AGL-101 · Pull the answer out of a model message

agent-loop · easy · implement · ~8 min · no AWS account

A model's final message is not a string. It is a list of content blocks, and only some of them carry text. Every agent loop has a line that turns that list into the sentence a user reads — and it is a line people get subtly wrong for months.

The shape

{"role": "assistant",
 "content": [{"text": "Let me check that booking."},
             {"toolUse": {"toolUseId": "t1", "name": "get_booking", "input": {}}},
             {"text": "It was cancelled by the carrier, so it is refundable."}]}

Three blocks. Two are text. One is a tool request the loop has already handled. The answer is the two text blocks, in order, joined — not the last block, and not the first.

Implement

def answer_text(message: dict) -> str:
    """Return the user-facing text of an assistant message.

    - Join every text block, in order, with a single space.
    - Ignore non-text blocks (toolUse, image, document …).
    - Strip leading/trailing whitespace from the result.
    - A message with no text blocks returns "" — never None, never raises.
    """
    # TODO

Copy the block, fill it in, and post it as a comment:

/drill AGL-101

```python
def answer_text(message):
    ...
```

What this proves

That you read the message shape rather than assuming the answer is content[0]["text"] or content[-1]["text"]. Both assumptions pass a demo and fail the first time the model narrates before calling a tool.

---

Submit in this thread

Copy the block from the brief, edit it, and post it as a comment here. The bot replies within a couple of minutes with every check, the skill you demonstrated, and what to do next.

_{Brief in the repo: labs/drills/agent-loop/AGL-101 · next in the sequence: AGL-102 · how grading works · scoreboard}