business-knowledge-interview
After a database is discovered and profiled, interview the business owner to capture the meaning-critical decisions the rest of the Database-to-PBIP flow depends on -- KPI meaning, PII handling, table grain, primary keys, relationships, missing-value rules, and ambiguous financial/quantity/date columns -- and record every outcome in the project Decision Store. Use when someone asks to run the interview, capture business knowledge, or resolve the business questions behind a model in the Seshat BI repo. It asks obvious low-risk items as one batch and critical items individually with explicit named-human approval; it uses masked samples and never shows raw suspected PII by default; it records decisions and STOPS. It does NOT define metric contracts or DAX, does NOT self-grant any approval, and does NOT emit a confidence score.
- License 许可
- Apache-2.0
- 最近更新
- 2026-11-11
- Stars
- ★ 2
- Forks
- 0
01 前言 Front Matter
- name
- business-knowledge-interview
- description
- After a database is discovered and profiled, interview the business owner to capture the meaning-critical decisions the rest of the Database-to-PBIP flow depends on -- KPI meaning, PII handling, table grain, primary keys, relationships, missing-value rules, and ambiguous financial/quantity/date columns -- and record every outcome in the project Decision Store. Use when someone asks to run the interview, capture business knowledge, or resolve the business questions behind a model in the Seshat BI repo. It asks obvious low-risk items as one batch and critical items individually with explicit named-human approval; it uses masked samples and never shows raw suspected PII by default; it records decisions and STOPS. It does NOT define metric contracts or DAX, does NOT self-grant any approval, and does NOT emit a confidence score.
03 SKILL.md 正文
business-knowledge-interview
The interview is the front door of the Database-to-PBIP flow's decision layer (spec 121). It converts a discovery profile into recorded, approvable business decisions so that every downstream stage consumes governed answers instead of unrecorded assumptions.
Behavior is governed by the business-knowledge-interview contract and its
recorded interview protocol. The
static rules DS1-DS5 (seshat check) verify the recorded outcome, not the
conversation.
When to use
- After
retail-onboard-tablehas produced a committed discovery profile (or an equivalent committed profile exists). - Before KPI contracts, Silver/Gold planning, DAX, dashboard blueprint, or PBIP readiness -- those stages are blocked until their critical decisions are approved.
Preconditions (stop if unmet)
- A committed discovery profile exists. No profile => do not start; report the gate as blocked naming the missing profile.
- Load the existing Decision Store first. Existing decisions are presented for confirmation or supersession -- never overwritten.
How it runs
- Ground every question in the profile (profile summaries, candidate grains, column types). Do not walk every column one-by-one unless the owner asks or an ambiguity requires it. Concentrate on: KPI inputs, PII, table grain, keys, relationships, missing-value rules, and ambiguous financial/quantity/date columns.
- Batch the obvious low-risk items into one owner confirmation. Critical
decision types are never in a batch. The owner may exclude any item; the
remainder is approved in one action and each excluded item becomes an
individual
pendingquestion. - Ask critical decisions individually with an explicit per-decision approval
by a named human whose authority class is eligible for the decision type
(as recorded in the
approval-authoritycontract). Route any KPI-meaning question toretail-kpi-knowledge; never invent meaning here. - Ask the domain's owner questions -- only those the scoped KPIs depend on. For
each retail KPI domain the table feeds, open that domain pack in the
retail-kpi-knowledgeskill and ask the cards in its## Owner questionssection that bear on a KPI actually in scope. Do not ask a card no scoped KPI depends on: an unanswered card is recorded as a criticalpendingdecision, and the decision gate filters by decision type, not by KPI, so an out-of-scope card blocks the whole stage (asking for Net Sales by Branch must not leave a same-store ruling pending). If scope later widens, ask the cards it newly depends on then. Each card names thedecision_typeits answer records under. A card's stated layer default is CONTEXT shown to the owner -- never record it as their ruling. An unanswered in-scope card is recordedpending/needs_user_input; this layer never invents a policy to make a number appear (that layer's ambiguity Resolution rule). Not every card type blocks: thekpi_contractsgate blocks onkpi_definition,policy_rulingandmissing_value_ruleonly, so a pendingdata_exclusionortable_graincard is recorded but does NOT stop contract authoring. Report those to the owner explicitly as open-but-not-blocking rather than implying the gate will catch them -- the gate's category list is the authority, not this step. This step supplies the KPI-meaning questions step 3 routes out; it does not define a metric or a contract. - Mask suspected PII by default. Show shape-preserving masks, cite the
suspicion source. Unmasking requires an explicit owner instruction, recorded
as a
pii_handlingdecision scoped to the affected columns. Never write a raw suspected-PII value into a committed store file. - Record every outcome in the Decision Store
(
.seshat/semantic-decisions.yaml,.seshat/kpi-contracts.yaml,.seshat/cleaning-rules.yaml) -- answered =>proposed/approved; refused =>rejected/deferred; unanswered =>pending/needs_user_input; sample needed but unavailable =>needs_sample. Confidence (low/medium/high) is the agent's proposal confidence only; it is never approval and never a readiness signal.
Exit
Regenerate evidence/business-interview-review.md from the store and report the
current gate verdict for the next requested stage.
Hard stops (never)
- Never self-grant an approval (Principle V;
never_self_grant_approval). - Never advance a readiness stage.
- Never emit a numeric confidence score (
never_fabricate_a_confidence_score). - Never display raw suspected-PII values by default, and never commit them.
- Never define a metric contract or DAX (that is the retail-kpi / DAX layers).
04 同仓库更多 Skills
fake-skill
A fixture skill with declaring frontmatter, used as positive shipped evidence.
fake-skill
A fixture skill with declaring frontmatter, used as positive shipped evidence.
bare-skill-no-frontmatter
暂无简介
fake-skill
A fixture skill with declaring frontmatter, used as positive shipped evidence.
retail-kpi-knowledge
Retail KPI / metric-contract reasoning layer for BI agents in the Seshat BI project. Use when an agent must reason about the BUSINESS MEANING of a retail KPI before any implementation — defining a KPI in business terms, classifying additivity (fully/semi/non-additive), identifying grain, listing required source fields, resolving ambiguity (gross vs net, VAT, returns, cost method, same-store), choosing an MVP KPI pack, reviewing a metric contract for completeness, or preparing a clean handoff to an implementation layer. Also use it to classify KPI sufficiency or prepare an owner policy decision packet. This is a meaning/definition + review layer, not an executor: it does not write DAX/SQL/Python, grant readiness, approve policy, or design dashboards. See INDEX.md for live/planned routes; registry.yaml is the authoritative machine-readable inventory.
bi-sql-knowledge
SQL reasoning, validation, reconciliation, transformation, diagnostics, and anti-pattern knowledge layer for Seshat BI agents. Use when a task involves reasoning about source SQL: source profiling, table grain, keys & uniqueness, joins & fan-out amplification, aggregation correctness, COUNT/NULL semantics, deduplication, validation queries, reconciliation queries, silver/gold transformation logic, window/date-time analytics, SQL anti-patterns, or basic SQL performance reasoning, or interpretation of a supplied sanitized PostgreSQL JSON execution plan. The SQL counterpart to bi-dax-knowledge. It is a thinking & validation layer, not a database executor, runtime validator, dbt/Dagster project, or automatic tuning service.