🔬 Research
dbt Test Severity Checklist from Manifest Notes (No Invented Pass Rates)
Compile a dbt test-severity checklist from pasted manifest notes only. No invented pass rates, row counts, or freshness lag scores.
0Reviews
Prompt
Act as a dbt analytics engineer who only uses pasted manifest notes. You compile a test-severity checklist the notes already support. You do not invent pass rates, row counts, freshness lag scores, or warehouse cost ranks. This is not a live dbt test run and not a data-quality SLA certificate. You work only from Inputs. Do not invent stats, citations, quotes, URLs, names, IDs, or records that are not in Inputs. Inputs: - Manifest notes I lock (model stubs, test cues, severity cues): [ManifestNotes] - dbt version or adapter notes I lock: [Version] - Project or package label I may quote (or UNKNOWN): [ProjectLabel] - Model names already present (or UNKNOWN): [ModelNames] - Test names already present (or UNKNOWN): [TestNames] - Severity labels already present (or UNKNOWN): [SeverityLabels] - Source or ref cues already present (or UNKNOWN): [SourceCues] - Words I must not use: [Banned] - What I must never invent (pass rates, row counts, freshness lag scores, warehouse cost ranks): [Never] - Output format: [Format] - Language: [Lang] Generate: 1. Honesty ledger: ManifestNotes nouns, Version, ProjectLabel, ModelNames, TestNames, SeverityLabels, SourceCues, Lang. Forbidden: invented pass rates, row counts, freshness lag scores, warehouse cost ranks. 2. Test-severity checklist: one checkbox row per ModelNames entry. Attach only TestNames named beside that model in ManifestNotes. Missing test write NOT IN INPUTS. 3. Severity sketch: for each SeverityLabels entry, list tests that name it. Do not invent a 99.2% pass rate if absent. 4. Source caution block: quote SourceCues only. Freshness SLA packs not in ManifestNotes stay NOT IN INPUTS. 5. Refuse list: inventing 99.2% pass rates, inventing 4.2M row counts, inventing 18-minute freshness lag, inventing warehouse cost rank #1. 6. Compliance pass: quote Banned and Never hits. Cut them. Print model and test counts from ManifestNotes only. Format as Format. Constraints: - Test-severity checklist from ManifestNotes only. No invented pass rates. - Honor Version. No emojis. Not a live dbt Cloud run history.
Instructions
Replace every [bracket] with your details before running. Works on ChatGPT, Claude, and Gemini.
Generated Output
This image was generated using the prompt above.
Examples
Example Input
ManifestNotes: model fct_harbor_orders as pasted test not_null_order_id as pasted; model dim_quay_customers as pasted test unique_customer_id as pasted. Pass invent NONE. Rows invent NONE. Version: dbt Core 1.8 as pasted (do not invent unreleased AI quality coach) ProjectLabel: Harbor Quay analytics as pasted ModelNames: fct_harbor_orders as pasted; dim_quay_customers as pasted. Cedar staging UNKNOWN. TestNames: not_null_order_id as pasted; unique_customer_id as pasted. Midpoint accepted_values UNKNOWN. SeverityLabels: severity warn as pasted for not_null_order_id; unique_customer_id severity NOT IN INPUTS. SourceCues: source cue raw_harbor_orders as pasted. Freshness SLA UNKNOWN. Banned: 99.2% pass, 4.2M rows, guaranteed 18-minute freshness Never: invent pass rates, invent row counts, invent freshness lag scores, invent warehouse cost ranks Format: ledger + test-severity checklist + severity sketch + source caution + refuse + compliance Lang: English
Example Output
1. Ledger. ManifestNotes: model fct_harbor_orders + test not_null_order_id; model dim_quay_customers + test unique_customer_id. Pass invent NONE. Rows invent NONE. Version dbt Core 1.8. ProjectLabel Harbor Quay analytics. ModelNames fct_harbor_orders; dim_quay_customers. TestNames not_null_order_id; unique_customer_id. SeverityLabels warn for not_null_order_id; unique_customer_id severity NOT IN INPUTS. SourceCues raw_harbor_orders. Freshness SLA UNKNOWN. Lang English. Forbidden: invented pass rates, row counts, freshness lag scores, warehouse cost ranks, 99.2% pass, 4.2M rows, guaranteed 18-minute freshness. 2. Test-severity checklist. [ ] fct_harbor_orders | test not_null_order_id as pasted. [ ] dim_quay_customers | test unique_customer_id as pasted. Midpoint accepted_values not attached. Cedar staging not added. 3. Severity sketch. severity warn | test not_null_order_id as pasted. unique_customer_id severity | NOT IN INPUTS. Pass rate NOT IN INPUTS so do not invent 99.2%. Second warn cue not invented. 4. Source caution. raw_harbor_orders as pasted. Freshness SLA UNKNOWN. Warehouse cost pack NOT IN INPUTS. Do not invent freshness lag packs. 5. Refuse. 99.2% pass rates: refused. 4.2M row counts: refused. 18-minute freshness lag: refused. warehouse cost rank #1: refused. Unreleased AI quality coach: refused. 6. Compliance. Banned hits none. Models 2. Tests 2. Format ledger+test-severity checklist+severity sketch+source caution+refuse+compliance. Gaps: unique_customer_id severity, Cedar staging, Midpoint accepted_values, Freshness SLA, pass rate. Missing-data policy: if a field was blank, write NOT IN INPUTS rather than guessing. Lock any tool version named in Inputs; if unnamed, write unknown. No invented testimonials, star ratings, or press logos. If legal, clinical, insurance, HR, education-plan, or veterinary content appears, add a one-line not-advice and de-identify banner. Quote banned-word hits and cut them. End with a gaps list of five bullets the user still owes you. Character and byte caps in the job are hard; print counts when relevant. Refuse to backfill DOIs, exam dumps, PHI, PII, or compensation promises not in Inputs.