From signalpilot-dbt
Load when task involves adding, writing, or fixing dbt unit tests. Covers unit_tests YAML format, given/expect blocks, edge-case coverage, and the difference between unit tests and schema tests.
How this skill is triggered — by the user, by Claude, or both
Slash command
/signalpilot-dbt:dbt-testingThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Schema tests (`unique`, `not_null`, `accepted_range`) check constraints on output data.
Schema tests (unique, not_null, accepted_range) check constraints on output data.
Unit tests check computation logic with mock inputs and expected outputs.
When a task says "add tests," determine which type:
If the model already has schema tests and the task says "add tests," it means unit tests.
Do NOT add schema tests when the task asks to verify logic - schema tests cannot detect formula bugs like reversed signs, wrong denominators, or incorrect window sizes.
Unit tests live in a .yml file next to the model. Top-level key is unit_tests:, NOT inside models:.
unit_tests:
- name: test_order_total_basic
model: my_orders
given:
- input: ref('stg_orders')
rows:
- {order_id: 1, quantity: 3, unit_price: 10.00}
- {order_id: 2, quantity: 1, unit_price: 25.00}
- input: ref('stg_discounts')
rows:
- {order_id: 1, discount_amount: 5.00}
expect:
rows:
- {order_id: 1, total: 25.00}
- {order_id: 2, total: 25.00}
Rules:
given lists every ref() the model uses. Each entry has input and rows.expect.rows uses the model's output column names from its SELECT.given rows must match the input model's column names exactly.expect rows must match the tested model's output column names exactly.expect rows - only listed columns are checked.Before writing any unit test, read the model's SQL file. Identify:
ref() calls - these are the inputs for given blocks.expect rows.Do NOT guess input or output column names. Read the SQL.
When the task says "add tests," add tests. Do NOT rewrite, refactor, or rename columns in the model SQL. Unit tests verify existing logic - they do not change it.
If a unit test reveals a bug in the model (e.g., integer division truncation), document the finding. Do NOT delete the failing test to make the suite green. A failing test that catches a real bug is correct.
Write tests that isolate specific behaviors. Each test should catch exactly one class of bug.
Required edge cases for numeric/aggregation models:
For rolling window models, add:
For ratio/percentage models, add:
Name tests descriptively: test_<model>_<behavior> (e.g., test_daily_nps_all_negative_reviews).
# Run unit tests for a specific model
dbt test --select <model_name>,test_type:unit
# Run all unit tests in the project
dbt test --select test_type:unit
After writing unit tests, run them. If tests fail against the current model, that means the tests caught a bug. Report the finding. Do NOT delete the test.
Put unit test YAML in the same directory as the model's existing .yml file. If the model's schema tests are in models/marts/agg.yml, add unit tests to models/marts/agg.yml (or a new file like models/marts/unit_tests.yml). Either location works - dbt finds unit tests by scanning all .yml files.
unit_tests: inside a models: block - it is a top-level key.npx claudepluginhub signalpilot-labs/codex-signalpilot-plugin --plugin signalpilot-dbtSearches MemPalace before answering questions about past work, people, projects, or prior decisions. Returns verbatim stored content instead of guessing from model memory.
Guides Payload CMS config (payload.config.ts), collections, fields, hooks, access control, APIs. Debugs validation errors, security, relationships, queries, transactions, hook behavior.
Implements vector databases with Pinecone, Weaviate, Qdrant, Milvus, pgvector for semantic search, RAG, recommendations, and similarity systems. Optimizes embeddings, indexing, and hybrid search.