v0.10.0 — BigQuery MLOps Skill, Multi-Skill Composition Playbooks & Unified 3-Stage Architecture Visuals

[0.10.0] - 2026-10-26

Added

• 4th Bundled Skill (bigquery-mlops): Added skills/bigquery-mlops/SKILL.md and four Lens 2 operational reference guides (orchestration-and-pipelines.md, model-registry-export-and-serving.md, monitoring-drift-and-evaluation.md, and feature-store-and-temporal-joins.md) operationalizing ML.* (bigquery-ml), AI.* (bigquery-ai), and external models logging to BigQuery across all 8 orchestration engines, Vertex AI Model Registry (model_registry='VERTEX_AI'), EXPORT MODEL (xgboost_version='2.1') / IMPORT MODEL (ONNX/XGBOOST/TENSORFLOW), REMOTE WITH CONNECTION endpoints, TRANSFORM_ONLY preprocessing models, table-native drift/skew monitoring (ML.VALIDATE_DATA_DRIFT, ML.VALIDATE_DATA_SKEW, ML.DESCRIBE_DATA), model-free ML.METRICS, and point-in-time feature stores (ML.ENTITY_FEATURES_AT_TIME, ML.FEATURES_AT_TIME, and Vertex AI Feature Store).
• Multi-Skill Composition Playbooks & Problem-First Workflow Navigator (skills/bigquery-data-science/reference/cross-skill-playbooks.md): Added 6 end-to-end multi-skill architecture playbooks (Unstructured Featurization → Point-in-Time Join → Supervised Model, LLM Label Bootstrapping → Trained Student Model, Forecasting / Attribution / Causal Impact → Executive Narrative, Hybrid Search Retrieval → Trained Re-Ranker, External Model Logging to BigQuery → Universal Monitoring & Drift Brief, and Prototype Notebook → Production Scheduled Pipeline) plus a unified 7-stage problem-first index of all 32 workflows and 8 pipelines across bigquery-ml/ and bigquery-ai/.
• Sister-Skill Bridge Sections: Added Composing With Sister Skills tables to skills/bigquery-ml/SKILL.md, skills/bigquery-ai/SKILL.md, and skills/bigquery-mlops/SKILL.md.
• Multi-Skill Triage & MLOps Static SQL Review: Upgraded triage_bigquery_task in src/bigquery_data_science/catalog.py to return recommended_skills: list[str], added 2 new empirical comparisons (#9 Orchestration Engine Selection and #10 Serving, Registry & Portability), and added 4 new static SQL checks in review_bigquery_sql (model_registry='VERTEX_AI' vertex_ai_model_id, EXPORT MODEL xgboost_version='2.1', 3-argument ML.VALIDATE_DATA_DRIFT universal monitoring, and post-retrain useQueryCache=false).
• Unified 3-Stage Architecture Visuals: Added vector docs/assets/readme_architecture.svg and unified README.md Mermaid, PyPI docs/assets/readme_architecture.png, and 1280x640 docs/assets/social_card.png, plus upgraded all 7 repository Mermaid diagrams with explicit wrappingWidth and horizontal 3-card stage shelves.