Skip to main content

DDS Schema Reference

DDS.manifest.toml​

The standard manifest. Describes DDS itself: identity, governance relationships, artifact locations, and lifecycle.

[standard]
id = "dds"
title = "Dataset Development Standard"
abbreviation = "DDS"
status = "active"
maturity = "candidate"
version = "0.2.1"
description = "Governs datasets, corpora, provenance, licensing, validation, splits, hashes, usage notes, sensitive-data constraints, and release readiness."

[governance]
meta_standard = "SFDS"
workspace_standard = "WGS"
scope = "Datasets, training corpora, evaluation sets, reference archives, derived datasets, provenance, licensing, validation, splits, integrity, and usage constraints."
non_goals = ["Model training recipes", "website deployment", "desktop release packaging"]

[artifacts]
specification = "Dataset Development Standard.md"
schema = "DatasetManifest.schema.toml"
templates = "templates"
examples = "examples"
adoption_guide = "Adoption-Guide.md"
validation_checklist = "Validation-Checklist.md"
changelog = "CHANGELOG.md"
validators = []
governance_notes = []
reference_examples = [
  "examples/DDS-Suite-Map.md",
  "examples/Example-Provenance-Record.md",
  "examples/Example-License-Record.md",
  "examples/Example-Split-Record.md"
]

[adopter_artifacts]
schemas = ["DatasetManifest.schema.toml"]
manifest_templates = ["templates/DATASET.manifest.toml"]
document_templates = [
  "templates/Provenance-Record.md",
  "templates/License-Record.md",
  "templates/Split-Record.md"
]

[lifecycle]
created = "2026-06-10"
last_updated = "2026-06-11"
maintainer = "Herb"

[agent]
read_first = ["README.md", "Dataset Development Standard.md", "DDS.manifest.toml"]
notes = "Use DDS for local corpora, training data, and archive datasets."
  • [standard] — identity block: id, title, abbreviation, status, maturity, version, one-line description.
  • [governance] — links to SFDS (meta-standard) and WGS (workspace standard), states scope and non-goals.
  • [artifacts] — points to the specification, schema, templates/examples folders, adoption guide, validation checklist, changelog.
  • [adopter_artifacts] — schemas and templates that dataset projects actually consume: the manifest schema plus provenance, license, and split record templates.
  • [lifecycle] — creation/update dates and maintainer.
  • [agent] — reading order for agents onboarding onto DDS; notes it is used for local corpora, training data, and archive datasets.

DatasetManifest.schema.toml​

The schema that validates a project's DATASET.manifest.toml.

title = "DDS Dataset Manifest Schema"
type = "object"
required = ["dataset", "governance", "provenance", "license", "validation"]

[properties.dataset]
required = ["id", "title", "version", "class"]

[properties.governance]
required = ["standard", "workspace_standard"]

[properties.provenance]
required = ["sources", "created"]

[properties.license]
required = ["name", "usage_constraints"]

[properties.validation]
required = ["status", "checks"]
  • dataset — required identity fields: id, title, version, and dataset class (see the dataset classes table in the overview).
  • governance — required links back to the governing standard (DDS) and workspace standard (WGS).
  • provenance — required source list and creation date.
  • license — required license name and usage constraints.
  • validation — required status and list of checks performed.