---
type: Dataset
id: https://w3id.org/lmodel/slsa/knowledge/datasets/test-data
genre: reference
title: Test data
description: Example YAML instances of the slsa schema in tests/data/, valid and invalid, named ClassName-###.yaml after the class they instantiate (SlsaStatement, InTotoDigestSet, SlsaBuildPlatform and the rest); pytest and linkml-run-examples check them.
resource: tests/data/
sources:
  - resource: tests/data/README.md
    id: data-readme.md
    title: the naming rule and the three subfolders
  - resource: tests/data/
    id: data
    title: the example files
  - resource: tests/test_data.py
    title: the pytest data test
  - resource: tests/test_schema_rules.py
    title: the checks of rules the schema states twice
  - resource: justfile
    title: the _test-examples recipe
  - resource: CONTRIBUTING.md
    title: the naming rule for example data
  - resource: src/slsa/schema/slsa.yaml
    title: what a SlsaDocument holds
  - resource: src/slsa/schema/slsa_core.yaml
    title: the subclasses of SlsaParty
dependsOn:
  - https://w3id.org/lmodel/slsa/knowledge/datasets/slsa-linkml-schema
generated:
  by: process:ktl-librarian
  at: "2026-10-30T23:13:40Z"
status: draft
verified:
  - by: process:ktl-librarian
    at: "2026-10-23T20:08:09Z"
---

# Test data

`tests/data/README.md` says that each file is named `ClassName-###.yaml`, where the part before the first `-` is the name of a class as the schema writes it, and sorts the files into three subfolders.[^data-readme.md]

- `valid/` holds 32 files[^data] of data that conforms to the schema:
  - statements for each kind of predicate: `SlsaStatement` files for build provenance and Tekton build provenance, and `InTotoStatement` files for a verification summary, source provenance, source verification summaries for a git and an SVN revision, and a build environment attestation;
  - files for `InTotoDigestSet`, `InTotoResourceDescriptor` and `SlsaDocument`, and for the build provenance and verification summary predicates on their own;
  - a file for each role a party can act in, which the schema models as the subclasses of `SlsaParty`: `SlsaProducer`, `SlsaConsumer`, `SlsaVerifier`, `SlsaInfrastructureProvider` and `SlsaBuildImageProducer`;
  - files for the other `Slsa` classes a document holds: the build platform and its control plane, the package and its registry, the source control system, repository and revision, a branch and a tag, the build image, the dependency inventory and the adoption metadata.
- `invalid/` holds 36 files of data that must fail, and the first line of each says why. Among them are:
  - an `InTotoStatement` with no subject, a subject with no digest, the wrong statement type or a predicate of the wrong kind;
  - an `InTotoDigestSet` that is empty or whose SHA-256 digest is in uppercase hexadecimal;
  - an `InTotoResourceDescriptor` with no URI, digest or content;
  - a `SlsaBuildPlatform` with an unknown level;
  - a `SlsaBuildProvenance` whose start time carries a `+01:00` offset, where an in-toto timestamp ends in Z;
  - a `SlsaVerificationSummary` with an unknown result, a policy with no URI, or a verifier id whose host is in uppercase.
- `problem/` holds data the current schema does not handle correctly yet, split into `valid/` and `invalid/`. Both are empty today.

`tests/test_data.py` loads every file in `valid/` with the generated dataclass model. `tests/test_schema_rules.py` checks rules the schema states in two places: the digest set's named algorithms against its at-least-one rule and its reserved names, and the `slsa_result_value` pattern against `SlsaResult`. `just test` also runs `linkml-run-examples` over `valid/` and `invalid/`, writing its output to the git-ignored `examples/output/`.

[^data-readme.md]: tests/data/README.md, the naming rule and the list of subfolders
[^data]: tests/data/, the YAML files in `valid/`, `invalid/` and `problem/`