Query: Add a few more example questions specific to CrateDB

CHANGES.md

@@ -4,6 +4,7 @@
 - Outline: Shrank llms-txt output to <200_000 input tokens
 - Bundle: Added "count-tokens" procedure to inform about the size
   of the outcome. Sonnet and Opus are limited to 200_000 input tokens.
+- Query: Added a few more example questions specific to CrateDB
 
 ## v0.0.7 - 2025-07-22
 - Prompt: Added `instructions-general.md` file when generating bundle

src/cratedb_about/cli.py

@@ -122,8 +122,8 @@ def ask(question: str, backend: t.Literal["claude", "openai"]) -> None:
     if not question:
         # Use the AUTOINCREMENT question or fall back to the first question if not found
         default_question = next(
-            (q for q in Example.questions if "AUTOINCREMENT" in q),
-            Example.questions[0] if Example.questions else "What is CrateDB?",
+            (q for q in Example.knowledgebase if "AUTOINCREMENT" in q),
+            Example.knowledgebase[0] if Example.knowledgebase else "What is CrateDB?",
         )
         question = default_question
     click.echo(f"Question: {question}\nAnswer:\n")
@@ -135,4 +135,21 @@ def list_questions() -> None:
     """
     List a few example questions about CrateDB.
     """
-    click.echo("\n".join(Example.questions))
+
+    output = f"""
+# Questions
+
+## Knowledgebase
+
+{Example.render(Example.knowledgebase)}
+
+## Text-to-SQL
+
+### time_series_data
+{Example.render(Example.text_to_sql["time_series_data"])}
+
+### summits
+{Example.render(Example.text_to_sql["summits"])}
+    """.strip()
+
+    print(output)  # noqa: T201

src/cratedb_about/query/model.py

@@ -2,19 +2,24 @@
 import os
 import typing as t
 from pathlib import Path
+from textwrap import dedent
 
 from cratedb_about.settings import settings
 from cratedb_about.util import get_cache_client
 
 logger = logging.getLogger(__name__)
 
 
+def cleanse(q: str) -> str:
+    return dedent(q).strip().replace("\n", " ")
+
+
 class Example:
     """
     A few example questions to ask about CrateDB.
     """
 
-    questions = [
+    knowledgebase = [
         "What are the benefits of CrateDB?",
         "Tell me about why CrateDB is different.",
         "Tell me about CrateDB Cloud.",
@@ -43,6 +48,32 @@ class Example:
         "How do I optimally synchronize data between MongoDB and CrateDB?",
     ]
 
+    text_to_sql = {
+        "time_series_data": [
+            "What is the average value for sensor 1?",
+            "What’s the name of the sensors table? Does it use any special cluster settings?",
+            "How many sensor measurements have been recorded?",
+            "When was the most recent measurement taken?",
+            "Do you see any irregularities in the recorded measurement values?",
+            cleanse("""
+            Can you plot the measurement values over time to visualize any trends or anomalies
+            by providing a Python code example and executing it?
+            Please highlight or annotate the outliers in the plot."""),
+            cleanse("""
+            Create several visualizations in Jupyter using matplotlib and seaborn based on data
+            in CrateDB’s `time_series_data` table, highlighting irregularities, and display them.
+            If you need to, please install required dependencies automatically."""),
+        ],
+        "summits": [
+            "What’s the highest mountains in Switzerland (CH) in the summits table?",
+            "Can you provide a list of the highest mountains by country from the dataset?",
+        ],
+    }
+
+    @classmethod
+    def render(cls, thing):
+        return "\n".join([f"- {item}" for item in thing])
+
 
 class KnowledgeContextLoader:
     """

tests/test_query.py

@@ -51,7 +51,7 @@ def test_example_question():
     """
     Validate the example question bundle class.
     """
-    assert "How to enumerate active jobs?" in Example.questions
+    assert "How to enumerate active jobs?" in Example.knowledgebase
 
 
 def test_ask_openai_no_api_key():