domain-model-lint

name	domain-model-lint
description	Run modelith lint on a domain model and explain the findings — a read-only review pass. Use when the user asks to check, validate, or review a *.modelith.yaml file, or wants to know what is missing or inconsistent. Turns the linter output into a prioritized, actionable summary. To actually change the model (add or edit entities, fix the gaps), use domain-model-author.

Linting a domain model

Run the linter and interpret it for the user. The point is not a clean bill of health — it's making gaps visible so the user can decide whether they matter.

Run it

First confirm modelith is installed (modelith --version); if it is not on PATH, tell the user to install it before going further — don't eyeball the YAML in its place:

go install github.com/stacklok/modelith/cmd/modelith@latest
# or download a binary from https://github.com/stacklok/modelith/releases

modelith lint <file>...            # human-readable
modelith lint --format json <file> # when you need to parse findings

Use --completeness error only if the user wants gaps to be treated as hard failures (e.g. a strict CI gate). It promotes the completeness warnings (the advisory gaps listed below) to blocking errors. It does not affect semantic errors — those (including a dangling invariants_touched id) block regardless of the flag.

Interpret the three layers

• Structural (error). The file violates the JSON Schema — wrong type, missing required field, bad cardinality — or names a cross-model reference (scope.Name) in relationship.entity or subtypeOf, which is not supported anywhere but an attribute type. Must be fixed; the model won't parse cleanly otherwise.
• Semantic (error or warning). Errors (always block, flag-independent): a relationship points at an entity that doesn't exist; a scenario's invariants_touched or an action's preserves names an invariant id no entity or top-level invariants entry declares; the same invariant id is declared twice (entity-level and top-level invariants share one id namespace); reciprocal cardinalities that aren't inverses; reciprocal declarations that both claim ownership: owned; an imports entry that's absolute, unreadable, not a domain model, binds a scope another entry already bound, resolves outside the repository holding this model, is a bare path whose filename yields no valid scope slug, has a path containing a control character, or declares a schema version this modelith doesn't support; an attribute type containing a dot that isn't a well-formed scope.Name (a dot is always reserved for a cross-model reference — decimal(10.2) and google.protobuf.Timestamp are errors, not primitives); a scope.Name type whose scope no import binds, or that names no enum in the model it resolves to. Warnings: a backticked term resolves to no entity, role, or actor; an action actor that's neither an entity nor a glossary term; a PascalCase attribute type that names no defined enum — usually a typo or a concept that was never named; a relationship role written as prose, which belongs in note; an ambiguous reciprocal pairing, where one end declares the same relationship twice and the other declares it back. Decide which it is and propose the fix.
• Completeness (advisory warning). Gaps, not bugs: an entity with no invariants, an entity no scenario exercises, a glossary term defined but never referenced, an enum no attribute uses, or an import nothing references. shared: true on the model being linted relaxes the two vocabulary-only checks (the unused glossary term and unused enum) — it does not relax the invariant or scenario-coverage checks, and it does nothing for an unreferenced import (that check is about this model's own imports, not about being imported). These are what --completeness error promotes to blocking. For each, ask whether it's a real gap (write the missing invariant or scenario, or drop the unused import) or genuinely fine — a shared: true model in particular.

How to report back

1. Lead with errors — these block. Give the fix for each.
2. Then semantic warnings — likely typos or missing definitions.
3. Then completeness gaps, framed as questions: "Nothing exercises Policy in a scenario — should there be one, or is it intentionally a supporting concept?"

Don't just dump the linter output. Prioritize, explain why each matters, and propose concrete edits. If the user agrees, use the domain-model-author skill to make them, then re-lint and re-render.