Drill · lap 1 · 3/9 · TOOL-101 · Fill in the honest empty result

[akash-coded]

3 of 9 · TOOL-101 · Fill in the honest empty result

tools · easy · blank · ~8 min · no AWS account

A policy search finds nothing and returns []. The model reads that and writes "there are no restrictions on this fare" — the opposite of what the tool meant. Fill in the four blanks so the return value cannot be misread.

def search_policy(query, corpus, index_ready=True):
    if not index_ready:
        return {"status": ____1____,            # the corpus was NOT searched
                "advice": "The policy corpus was not searched. Do not state what policy says; "
                          "tell the user you could not check and escalate."}

    matches = [p for p in corpus if query.lower() in p["text"].lower()]
    if not matches:
        return {"status": ____2____,            # searched, held nothing
                "searched_count": ____3____,    # evidence the search happened
                "advice": ____4____}            # what the model must NOT conclude
    return {"status": "ok", "matches": matches}

The blanks

	What goes here
____1____	A status string meaning the corpus was never searched
____2____	A status string meaning searched, and nothing matched — it must differ from blank 1
____3____	How many passages were searched
____4____	A sentence that forbids the wrong conclusion and names the next action

Post the completed function:

/drill TOOL-101

```python
def search_policy(query, corpus, index_ready=True):
    ...
```

What this proves

That your tools encode the difference between two absences the model reliably confuses. Whatever you do not put in the payload, the model cannot recover downstream.

---

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/tools/TOOL-101 · next in the sequence: RET-101 · how grading works · scoreboard}