Cassis does context maintenance for analytics agents. Your business definitions live in a Git repository you own, changes run against evals, and merging an approved pull request publishes the context your agents read over MCP.
getcassis.com · docs · blog
Three ways in, depending on whether you want to bootstrap context, see the full loop, or start from a clean ontology.
| ontology-bootstrap | Assemble a reviewable first ontology from the dbt project, schema, dashboards and docs you already have. Evidence stays attached and unknowns become questions. No account or Cassis key required. Read the measured write-up. |
| cassis-demo-stallora | See the whole loop on a sample marketplace warehouse. Fifteen minutes, from your own terminal and your own agent, nothing of yours involved. |
| cassis-ontology-starter | Start an ontology on your own schema. The CI gates, the MCP config and the modeling guide, with none of the content — that part is yours. |
The demo and starter are GitHub templates: use the template, don't fork.
| cassis-cli | Validate, test and eval an ontology from your terminal, then publish it. The same commands run in CI to gate pull requests. pip install cassis-cli |
| skills | Agent skills for authoring an ontology and for querying through MCP. /plugin marketplace add GetCassis/skills |
| cassis-ontology-examples | Two complete ontology trees to copy from: a minimal skeleton and a fully authored one. |
| dbt-agent-readiness | Audit a dbt project for what an AI agent will get wrong if you point it at the data today. Runs on your own repository; no Cassis account needed. |
| research | Benchmarks, harnesses and the run data behind our write-ups. |