Looker Ingest

Pulls runtime Looker dashboards, Looks, and explores from staged JSON and turns them into durable knowledge base entries and semantic layer sources. The workflow is strict: verify every table identifier with entity_details before writing, search existing wiki pages and SL sources to merge rather than duplicate, write candidates through context_candidate_write instead of direct wiki edits, and run sl_validate after every SL write. It handles the gnarly mapping between Looker's sql_table_name templates and actual warehouse tables through a targetTable contract, falling back to wiki-only mode with structured reason codes when the connection is unmapped or the SQL is too weird. If you're ingesting BI tool metadata into a knowledge graph, this is the reference implementation for doing it carefully.

GitHub stars: 1.1K
last updated: Jun 11, 2026

View on GitHub

Install

$ npx -y skills add kaelio/ktx --skill looker_ingest --agent claude-code

Installs into .claude/skills of the current project.

Files

Looker Runtime Ingest

Looker runtime ingest turns API-staged dashboards, Looks, and explores into durable ktx memory. Runtime entities are evidence. They are not themselves the final knowledge shape.