agentleFS
GetCassisGetCassis/ontology-bootstrap/SKILL.md

ontology-bootstrap

Skill31 starsChanged 14 days agoMITView on GitHub

Assemble a reviewable first version of an analytics agent's context from what a data stack already has: a warehouse schema export, a dbt project, docs, dashboards, query logs. Produces a Cassis ontology tree (domain READMEs, table and metric YAML, joins) with the evidence behind every claim attached and unresolved meaning filed as questions. Scripts do everything the inputs determine; the user decides at four checkpoints. Runs fully offline: no account, no key, no network. Optional dbt write-back merges descriptions into the project's own schema.yml files without overwriting anything already there. Use when asked to "bootstrap an ontology", "build context for an analytics agent", "document the warehouse for AI", "turn our dbt project into an ontology", or "make our data agent-ready".

  • Installs packages

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 1

    1. GetCassis/ontology-skills★ 29100% same

    Edited · 0

    None found.

    Wrote their own version · 1

    1. wlsdks/ontology-atlas★ 156same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository2 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use this entry point for a new context.

    line 20
  2. Very specific:

    Read CLAUDE.md in the checkout and follow it exactly.

    line 57

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Ontology bootstrap

This repository is both the skill and the tool: Python phases that do everything the inputs determine, plus an operating contract for the judgment stages.

Use this entry point for a new context. Once the team maintains the result, continue with skills/ontology-expansion/SKILL.md; rerunning the generator is not an incremental update to that maintained tree.

Before anything else

The run needs a full, writable checkout of this repository as its working directory, because every phase reads and writes inside it. If you are reading this file from an installed skill or plugin directory, do not run there. Clone the repository into a working directory and drive the run from the clone:

git clone https://github.com/GetCassis/ontology-skills
cd ontology-skills
python3 -m pip install -r requirements.txt

Installing this through a skill or plugin installer copies the files but does not give you a runnable kit: the copy is not a working directory, and the dependencies are not installed.

What it needs, and what it touches

Three pure-Python dependencies: pyyaml, sqlglot, ruamel.yaml. Nothing else, and no build step.

It runs fully offline. No account, no API key, no network call, and no connection to your warehouse — it reads a schema export you produce yourself. Where profiling would settle a question, the kit writes the SQL for a human to run and files the question; it never runs it.

It writes inside the checkout, in the run directory. The one exception is the optional --emit dbt, which merges descriptions back into your dbt project's own schema.yml files: it adds fields and never overwrites a description already there, and it round-trips the YAML so your comments and formatting survive. Everything dbt has no field for goes under meta.cassis.*.

python3 tests/test_kit.py runs on a fresh clone with no warehouse data and no credentials.

How to run

Read CLAUDE.md in the checkout and follow it exactly. It is the operating contract: the driver sequence (intake.py for a new run, then bootstrap.py prep, build, finish), the four checkpoints where you stop and wait for the user, the evidence and enrichment rules, and the reporting rules. Do not re-derive the sequence from the scripts.

README.md says what inputs the kit needs (a schema export is mandatory; the repository that builds the tables is strongly recommended, as a dbt project, a dbt docs export, or any other layout through a mapping config; everything else is optional) and what a run costs. Relay each phase's cost banner to the user before the phase spends it.

What comes out

<run>/emit/cassis/ is the result: domains as Markdown, tables, metrics and joins as YAML, with per-column provenance, ready for review like code. OUTPUT.md maps the rest of the run directory, including the open questions and the defect list the run surfaces along the way.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.