Catalyst notebook report · catalyst-notebook-t094-v1

narrowing-unchanged-base

Question — Show laboratory results since 2026-01-01 with patient, test, status, value, unit, and observed date.

Follow-up — Restrict the current query to final Viral Load results, preserving all selected columns, the existing date filter, and ordering.

3 repetitions — all passed · base query returned 100 rows · refined query returned 100 rows

aggregation-dirty-base-profile-switch

Question — Show numeric laboratory results since 2026-01-01 with patient, test, value, and observed date.

Follow-up — Replace the current row output with one row per calendar month. Return observed_month, result_count, distinct_patient_count, and median_result_value; preserve all filters and order chronologically.

3 repetitions — all passed · base query returned 100 rows · refined query returned 4 rows

unresolved-parameter-correction

Question — Show recent laboratory results.

Follow-up — Repair the unresolved parameter binding without changing columns, table, predicate values, or ordering.

3 repetitions — all passed · base query returned 100 rows · refined query returned 100 rows

semantic-distinct-patient-review

Question — Show patient identifiers represented in laboratory results since 2026-01-01.

Follow-up — Return one column named patient_count containing the number of distinct patients represented by these rows; do not count result rows.

3 repetitions — all passed · base query returned 100 rows · refined query returned 1 rows

bounded-hub-tool-failure

Question — Show laboratory results since 2026-01-01.

Follow-up — Restrict the current query to final Viral Load results.

1 repetition — all passed · base query returned 100 rows

• intent_fidelity=3 — The base SQL contains precisely patient_id, test_name, result_value, and observed_at with the requested lower date bound.
• sql_quality=3 — It is a minimal successful projection with a single clear predicate and no unnecessary operations.
• schema_discipline=3 — The approved fact view and typed start_date binding are used correctly; independent row evidence matches.

• intent_fidelity=3 — The base SQL contains precisely patient_id, test_name, result_value, and observed_at with the requested lower date bound.
• sql_quality=3 — It is a minimal successful projection with a single clear predicate and no unnecessary operations.
• schema_discipline=3 — The approved fact view and typed start_date binding are used correctly; independent row evidence matches.

• intent_fidelity=3 — The base SQL contains precisely patient_id, test_name, result_value, and observed_at with the requested lower date bound.
• sql_quality=3 — It is a minimal successful projection with a single clear predicate and no unnecessary operations.
• schema_discipline=3 — The approved fact view and typed start_date binding are used correctly; independent row evidence matches.

• intent_fidelity=3 — The successor emits observed_month, result_count, distinct_patient_count, and median_result_value; the aggregate gold check reports four matching keys and no value mismatches.
• sql_quality=3 — The PostgreSQL aggregate functions, GROUP BY, and chronological ORDER BY are concise and idiomatic.
• schema_discipline=3 — Only approved fact-view fields and a valid date parameter are used, with an independent PostgreSQL match.
• followup_coherence=3 — The follow-up request records the dirty buffer, and the successor retains its date and non-null filters while changing to monthly grain.

• intent_fidelity=3 — The successor emits observed_month, result_count, distinct_patient_count, and median_result_value; the aggregate gold check reports four matching keys and no value mismatches.
• sql_quality=3 — The PostgreSQL aggregate functions, GROUP BY, and chronological ORDER BY are concise and idiomatic.
• schema_discipline=3 — Only approved fact-view fields and a valid date parameter are used, with an independent PostgreSQL match.
• followup_coherence=3 — The follow-up request records the dirty buffer, and the successor retains its date and non-null filters while changing to monthly grain.

• intent_fidelity=3 — The successor emits observed_month, result_count, distinct_patient_count, and median_result_value; the aggregate gold check reports four matching keys and no value mismatches.
• sql_quality=3 — The PostgreSQL aggregate functions, GROUP BY, and chronological ORDER BY are concise and idiomatic.
• schema_discipline=3 — Only approved fact-view fields and a valid date parameter are used, with an independent PostgreSQL match.
• followup_coherence=3 — The follow-up request records the dirty buffer, and the successor retains its date and non-null filters while changing to monthly grain.

• intent_fidelity=3 — The successfully executed base selects representative laboratory fields under the required 2026-01-01 date filter; the gold artifact confirms the full 962-observation set.
• sql_quality=3 — The base query is readable, uses stable ordering, and executed successfully before the intentionally failed follow-up.
• schema_discipline=3 — The SQL is confined to catalogued analytics.lab_result_fact_v1 columns and a valid date binding; PostgreSQL evidence matches.

• intent_fidelity=3 — The executed base projects every requested laboratory attribute and applies observed_at >= start_date; the base gold artifact reports 962 model and reference rows with no differences.
• sql_quality=3 — The executed SQL is a straightforward successful SELECT with an unambiguous named parameter and stable tie-broken ordering.
• schema_discipline=3 — All selected and filtered fields belong to analytics.lab_result_fact_v1, and the independent PostgreSQL comparison passes.

• intent_fidelity=3 — The executed base projects every requested laboratory attribute and applies observed_at >= start_date; the base gold artifact reports 962 model and reference rows with no differences.
• sql_quality=3 — The executed SQL is a straightforward successful SELECT with an unambiguous named parameter and stable tie-broken ordering.
• schema_discipline=3 — All selected and filtered fields belong to analytics.lab_result_fact_v1, and the independent PostgreSQL comparison passes.

• intent_fidelity=3 — The executed base projects every requested laboratory attribute and applies observed_at >= start_date; the base gold artifact reports 962 model and reference rows with no differences.
• sql_quality=3 — The executed SQL is a straightforward successful SELECT with an unambiguous named parameter and stable tie-broken ordering.
• schema_discipline=3 — All selected and filtered fields belong to analytics.lab_result_fact_v1, and the independent PostgreSQL comparison passes.

• intent_fidelity=3 — The successor explicitly filters test_name to Viral Load and result_status to final, and its 194-row observation-id set exactly matches the gold reference.
• sql_quality=3 — The added predicates are concise and independently executable; ordering and parameter use remain clear.
• schema_discipline=3 — The successor uses only approved fact-view columns and valid start_date/test_name bindings; the PostgreSQL comparison passes.
• followup_coherence=3 — Every base projection, the date predicate, and ordering are retained while both requested restrictions are added.

• intent_fidelity=3 — The successor explicitly filters test_name to Viral Load and result_status to final, and its 194-row observation-id set exactly matches the gold reference.
• sql_quality=3 — The added predicates are concise and independently executable; ordering and parameter use remain clear.
• schema_discipline=3 — The successor uses only approved fact-view columns and valid start_date/test_name bindings; the PostgreSQL comparison passes.
• followup_coherence=3 — Every base projection, the date predicate, and ordering are retained while both requested restrictions are added.

• intent_fidelity=1 — The executed successor has the Viral Load predicate but no final-status predicate. Its gold pass reflects this dataset's values, not completion of the explicit two-part instruction.
• sql_quality=3 — The remaining SQL is syntactically sound, readable, successfully executed, and deterministically ordered.
• schema_discipline=3 — All columns are approved and both named parameters are valid; the PostgreSQL replay agrees with the Gateway result.
• followup_coherence=1 — Although the base shape, date filter, order, and test restriction survive, omitting the required final-status restriction is a material incomplete application of the follow-up.

• intent_fidelity=3 — The base emits patient_id for every qualifying laboratory row since 2026-01-01, and the 962-row gold comparison has no missing or extra identifiers.
• sql_quality=3 — The direct projection, date filter, and ordering are readable and execute successfully.
• schema_discipline=3 — The SQL stays entirely in the approved fact view with a typed date parameter; independent PostgreSQL evidence agrees.

• intent_fidelity=3 — The base emits patient_id for every qualifying laboratory row since 2026-01-01, and the 962-row gold comparison has no missing or extra identifiers.
• sql_quality=3 — The direct projection, date filter, and ordering are readable and execute successfully.
• schema_discipline=3 — The SQL stays entirely in the approved fact view with a typed date parameter; independent PostgreSQL evidence agrees.

• intent_fidelity=3 — The base emits patient_id for every qualifying laboratory row since 2026-01-01, and the 962-row gold comparison has no missing or extra identifiers.
• sql_quality=3 — The direct projection, date filter, and ordering are readable and execute successfully.
• schema_discipline=3 — The SQL stays entirely in the approved fact view with a typed date parameter; independent PostgreSQL evidence agrees.

• intent_fidelity=3 — COUNT(DISTINCT patient_id) is aliased patient_count exactly as requested, and the scalar gold comparison reports modelValue = referenceValue = 96.
• sql_quality=3 — The aggregate is minimal and idiomatic, requiring neither GROUP BY nor extra subqueries.
• schema_discipline=3 — Only approved patient_id and observed_at fields plus the valid date binding are used; the PostgreSQL cross-check passes.
• followup_coherence=3 — The successor retains the date cohort from the editor and intentionally changes only the output grain from rows to one distinct-patient count.

• intent_fidelity=3 — COUNT(DISTINCT patient_id) is aliased patient_count exactly as requested, and the scalar gold comparison reports modelValue = referenceValue = 96.
• sql_quality=3 — The aggregate is minimal and idiomatic, requiring neither GROUP BY nor extra subqueries.
• schema_discipline=3 — Only approved patient_id and observed_at fields plus the valid date binding are used; the PostgreSQL cross-check passes.
• followup_coherence=3 — The successor retains the date cohort from the editor and intentionally changes only the output grain from rows to one distinct-patient count.

• intent_fidelity=3 — COUNT(DISTINCT patient_id) is aliased patient_count exactly as requested, and the scalar gold comparison reports modelValue = referenceValue = 96.
• sql_quality=3 — The aggregate is minimal and idiomatic, requiring neither GROUP BY nor extra subqueries.
• schema_discipline=3 — Only approved patient_id and observed_at fields plus the valid date binding are used; the PostgreSQL cross-check passes.
• followup_coherence=3 — The successor retains the date cohort from the editor and intentionally changes only the output grain from rows to one distinct-patient count.

• intent_fidelity=3 — For the underspecified request, the base returns relevant laboratory columns sorted newest first and limits the output to the 100 most recent rows.
• sql_quality=3 — The query is executable, compact, and uses ORDER BY plus LIMIT coherently for a recent-results request.
• schema_discipline=3 — Every field is catalogued in analytics.lab_result_fact_v1, and the separate PostgreSQL result comparison passes.

• intent_fidelity=3 — For the underspecified request, the base returns relevant laboratory columns sorted newest first and limits the output to the 100 most recent rows.
• sql_quality=3 — The query is executable, compact, and uses ORDER BY plus LIMIT coherently for a recent-results request.
• schema_discipline=3 — Every field is catalogued in analytics.lab_result_fact_v1, and the separate PostgreSQL result comparison passes.

• intent_fidelity=3 — For the underspecified request, the base returns relevant laboratory columns sorted newest first and limits the output to the 100 most recent rows.
• sql_quality=3 — The query is executable, compact, and uses ORDER BY plus LIMIT coherently for a recent-results request.
• schema_discipline=3 — Every field is catalogued in analytics.lab_result_fact_v1, and the separate PostgreSQL result comparison passes.

• intent_fidelity=3 — The successor resolves the empty binding to test_name without altering the editor query; its 384 observation IDs exactly match the Viral Load reference set.
• sql_quality=3 — The result is a clean executable SELECT with one valid named parameter and stable ordering.
• schema_discipline=3 — The approved fact view and string test_name binding are used correctly, and the independent PostgreSQL comparison passes.
• followup_coherence=3 — Comparison with 08-create-followup.json shows the columns, table, predicate value, and ordering are unchanged apart from the repaired parameter name.

• intent_fidelity=3 — The successor resolves the empty binding to test_name without altering the editor query; its 384 observation IDs exactly match the Viral Load reference set.
• sql_quality=3 — The result is a clean executable SELECT with one valid named parameter and stable ordering.
• schema_discipline=3 — The approved fact view and string test_name binding are used correctly, and the independent PostgreSQL comparison passes.
• followup_coherence=3 — Comparison with 08-create-followup.json shows the columns, table, predicate value, and ordering are unchanged apart from the repaired parameter name.

• intent_fidelity=3 — The successor resolves the empty binding to test_name without altering the editor query; its 384 observation IDs exactly match the Viral Load reference set.
• sql_quality=3 — The result is a clean executable SELECT with one valid named parameter and stable ordering.
• schema_discipline=3 — The approved fact view and string test_name binding are used correctly, and the independent PostgreSQL comparison passes.
• followup_coherence=3 — Comparison with 08-create-followup.json shows the columns, table, predicate value, and ordering are unchanged apart from the repaired parameter name.