Isn't this just a toy?
No — and you don't have to take our word for it. The latest conformance run graded 473,312 cells of one rulebook's output across independent substrates — Postgres, Python, Go, TypeScript, Entity Framework, OWL — and every substrate with a closure engine agreed, cell for cell.
It's a fair question, because the examples are deliberately small enough to read in one sitting. But "small enough to read" and "toy" are different claims, and the second one is checkable.
What a toy would look like
A toy demo shows you one happy path on one stack and asks you to extrapolate. There is no answer key, no adversarial grading, and nothing stopping the demo from quietly special-casing whatever the slides need.
What this actually is
Every rulebook in the effortless-rulebooks repo is projected into independent substrates — Postgres, Python, Go, TypeScript, Entity Framework, OWL, Excel — and then graded row-by-row, column-by-column against a Postgres-oracle answer key. Not "the demo ran"; every derived cell in every substrate is compared. The latest run on the procedural-knowledge-ontology rulebook graded 473,312 cells at 99.96% — and the 0.04% shortfall is itself public and diagnosed (Excel has no closure engine yet, so its closure-derived cells fail honestly instead of being quietly skipped).
The content isn't toy content either
-
talismans-special-solutions is a faithful modeling of Jessica Talisman's Intentional Arrangement workflow-ontology series — role/agent separation, approval gates, delegation chains, PROV provenance, SKOS vocabularies — with its competency questions answered by graded, derived fields.
-
procedural-knowledge-ontology models a financial close whose step graph is cyclic (rework loops, fallback transitions), and derives answers like "which publications reached the end without ever passing a human approval gate?" — a real compliance question, answered by a declared closure over a filtered edge set.
-
This website, the fixed-bid delivery practice behind it, and the seeds you can clone are all driven by the same machinery.
The part that should actually move you
A typical Effortless project measures 60–80% derivative: run effortless -clean and that fraction of the codebase is deleted; run effortless build and it comes back, byte-for-byte, from the rulebook. Toys don't survive that operation. Systems whose single source of truth is real do.
So: the examples are small because legibility is the point. The grading is not.
The back story
Every answer here has a story behind it — and each story is (or becomes) a film.
A financial close that argues back
film not yet publishedThe procedural-knowledge-ontology rulebook models a monthly financial close the way one actually runs: steps that fall back to earlier steps, alternative paths that re-enter the flow, approval controls that can fire twice in one close. It was built across twenty-plus documented loops, each one adding witnesses — graded, derived fields that answer a compliance question the previous loop couldn't ask. The capstone witness, IsGateBypassedPublication, closes over only the transitions that avoid human approval gates and flags any publication reachable end-to-end without one. That is not a demo answering a planted question; it is a cyclic procedural ontology catching a control gap, identically, in every substrate it is projected into.
The librarian's ontology
film not yet publishedtalismans-special-solutions is a line-faithful modeling of Jessica Talisman's Intentional Arrangement series — a real information-science treatment of a production workflow, with role/agent separation, approval-gate subtypes, delegation chains, PROV provenance and SKOS vocabularies. The project's standing tie-breaker is written into its own docs: when something is ambiguous, the answer is never "what makes the badge green" — it is "what is most faithful to the article." Every competency question the series poses is answered by a graded, derived field, and every entity a question references is materialized in the UI. Faithfulness to someone else's published ontology is a test a toy cannot pass.