v0.3.5

[0.3.5] - 2026-06-25

Added

• New skills fabriciq-ontology-authoring-cli and fabriciq-ontology-consumption-cli — Fabric IQ Ontology (preview) support from the CLI. fabriciq-ontology-authoring-cli creates and evolves Ontology items (entity types, properties incl. timeseries, relationship types, and bindings to OneLake lakehouse or Eventhouse / KQL tables) via the Fabric item-definition REST API with a mandatory Preview & Confirm gate before any LRO write. fabriciq-ontology-consumption-cli reads Ontology items to produce agent grounding context and routes ontology-backed data queries by binding type to the matching per-datasource consumption skill (eventhouse-consumption-cli, spark-consumption-cli, sqldw-consumption-cli). Adds per-skill references/ (including a shared ontology schema reference bundled into each skill), routing tests, Vally integration evals, and full-eval plans.
• New skill: mlv-operations-cli -- Manage Materialized Lake View (MLV) refresh schedules and job execution via Fabric REST APIs. Provides scheduling and monitoring operations (9 endpoints):
  • Schedule Management: Create/list/update/delete refresh schedules (Cron, Daily, Weekly, Monthly)
  • Job Execution: Trigger on-demand refreshes, monitor job status/history, cancel running jobs
  • UX Patterns: Human-in-the-loop confirmations, step-by-step planning, iterative error handling
  • Gap Documentation: Transparently documents MLV discovery limitations — user must provide lakehouse ID and table names manually
• Vally integration tests -- 5 eval scenarios (2 with L1+L2 program verifiers, 3 with L0 output assertions + l1l2_exempt for error/edge cases)
• Cross-skill integration -- Routing from spark-authoring-cli, spark-operations-cli, FabricDataEngineer agent delegation
• Competitive advantage -- Fabric is first platform to offer conversational MLV scheduling (Databricks Lakeflow has no equivalent)