bigquery-sql
BigQuery-specific SQL patterns: UNNEST for array expansion, STRUCT, ARRAY_AGG,…
Financial reporting rules: grain consistency, balance sheets, double-entry ledgers, fiscal year boundaries, period-over-period calculations.
$ npx -y skills add signalpilot-labs/signalpilot --skill domain-financial --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/domain-financialContext preview
The summary Claude sees to decide when to auto-load this skill.
Financial reporting rules: grain consistency, balance sheets, double-entry ledgers, fiscal year boundaries, period-over-period calculations.
name: domain-financial description: "Financial reporting rules: grain consistency, balance sheets, double-entry ledgers, fiscal year boundaries, period-over-period calculations."
When aggregating transactions or metrics, drive FROM the fact/transaction table (journal_entries, invoices, payments), not the dimension table (accounts, customers). Accounts with zero transactions MUST NOT appear in transaction reports - they have no data to aggregate.
All metrics in a report MUST operate at the same grain - mixing grains produces silently wrong totals.
Use COUNT(*) on a line-item fact table, not COUNT(DISTINCT document_id) - the grain is the line item, not the document, so COUNT(*) is consistent with SUM(amount).
NEVER use COUNT(DISTINCT) just because a key represents a document or entity - a transaction fact table's grain is the transaction row.
Do NOT use a YML column description to choose between COUNT(*) and COUNT(DISTINCT) - descriptions explain semantics, not aggregation method. The source table's grain determines the correct aggregation.
A balance sheet has exactly three components: 1. Regular accounts (ASSET, LIABILITY, EQUITY) - cumulative running balances 2. Retained Earnings - cumulative P&L from ALL prior fiscal years 3. Current Year Earnings - cumulative P&L from the current fiscal year only
Query the project's configuration or organization table for fiscal year end settings - these values vary per tenant and MUST be used to compute the boundary, not hard-coded.
P&L transactions before the fiscal year boundary = Retained Earnings. P&L from the boundary onward = Current Year Earnings.
If the YML description specifies this earnings split, implement it exactly as written.
'Current fiscal year' means the fiscal year containing the run date, not the fiscal year of each reporting period, unless the description ties the boundary to the reporting period. On historical data this can leave the current-year component empty - that is correct, not a bug.
Every transaction has offsetting debits and credits - when unioning ledger entries, preserve BOTH sides.
Running balances MUST be computed with a window function ordered by `transaction_date` AND a deterministic tiebreaker (`transaction_id` or a sequence index) - ordering by date alone is non-deterministic when multiple transactions share a timestamp.
See dbt-workflow "Incremental Models and Period-Over-Period Columns" for the full rule. On first build, period-over-period columns MUST be NULL. Do NOT substitute 0.
Repo: signalpilot-labs/signalpilot
BigQuery-specific SQL patterns: UNNEST for array expansion, STRUCT, ARRAY_AGG,…
Load when dbt run or dbt parse fails. Covers YML duplicate patches, ref errors, passthrough…
Populate the knowledge base from dbt project research. Proposes entries across all 6…
Load when task involves dbt snapshots, SCD Type 2, or tracking data changes over time. Covers…
Load when task involves adding, writing, or fixing dbt unit tests. Covers unit_tests YAML…
Load when task involves dbt model versioning, creating v2 of a model, or backward-compatible…