lostinheaven-knt

chatbi-bootstrap

Procedural runbook for scaffolding a from-zero local Warehouse invoked by /chatbi-bootstrap. Writes local config (cli_adapters.mysql + optional path_binding), appends cli:mysql to shared adapters.query, CREATE DATABASE IF NOT EXISTS dw, introspects the source public schema via INFORMATION_SCHEMA, scaffolds the dbt-mysql project layout, and emits a source-inventory JSON hand-off for /chatbi-maintain-model. MySQL-only v1. Infra setup only (SEM-003). Carries reusable procedure, not easily-stale facts.

chatbi-build

Procedural runbook for deriving a build plan from a requirement, invoked by /chatbi-build-from-requirement. Reads blueprint (Source/Metrics/Layers/Tooling) + source_inventory + model_registry + semantic layer, derives an ordered ODS->DWD->DWS->ADS build plan (agent reasoning, no derivation lib), validates plan shape via chatbi_harness.build_plan, chains /chatbi-maintain-model per model, routes protected points to the human, and hands off to /chatbi-analyze. Orchestrator trust layer (SEM-003/META-008). Carries reusable procedure, not easily-stale facts.

chatbi-maintenance

Procedural runbook for governed model/semantic maintenance invoked by /chatbi-maintain-model. Generates a change-impact manifest (DOC-004), drives the model-metadata-semantic-reference-Skill-tests-downstream-eval sync gate, reuses the Cycle 3 Stop gate, and routes protected actions to the human owner (SEM-003). Carries reusable procedure, not easily-stale facts.

chatbi-runbook

Procedural runbook for the governed ChatBI analysis flow invoked by /chatbi-analyze. Concrete steps for clarification questions, evidence-backed tier degradation (T1 -> T2 -> T3), historical-SQL-as-clue handling, data-quality checks, candidate-SHA binding, independent adversarial-reviewer invocation, review/stop gate enforcement, and full provenance-footer assembly. It carries reusable procedure, not easily-stale facts.

chatbi-evaluation

Procedural runbook for governed offline evaluation invoked by /chatbi-evaluate. Enforces ground-truth isolation, seen/unseen separation, snapshot-anchored assertions, per-assertion scoring, EVAL-003 run records, owner-confirmed thresholds (EVAL-004), semantic-layer-use assertions (EVAL-005), and the FBK-003 "pass != absolute correctness" disclosure. Carries reusable procedure, not easily-stale facts.

chatbi-knowledge

Procedural runbook for authoring and maintaining governed knowledge references co-located with the Warehouse model. Enforces required metadata, "use for"/"do not use for" triggers, no machine absolute paths, historical-SQL-as-candidate-only, and cross-references (DOC-001/002/003/005). Carries reusable procedure, not easily-stale facts.