CDS v0.2.0 Specification

Status: Draft Issuer: SignedData.Org License: MIT Supersedes: CDS v0.1.0

1. Overview

The Curated Data Standard (CDS) is an open standard for distributing curated, cryptographically signed, real-time data from verified sources. Version 0.2.0 converts all identities — events, sources, content types — to dereferenceable HTTP URIs, making every CDS event valid JSON-LD. With this change, every CDS event becomes 5-star Linked Data: machine-readable, openly licensed, structured in a non-proprietary format, described using W3C standards, and linked to other people’s data via URIs.

Key properties of a CDS event remain unchanged:

• Typed — carries a content_type URI identifying the domain and schema
• Signed — RSA-PSS SHA-256 signature by the producer, verifiable by any consumer
• Fingerprinted — SHA-256 of the raw upstream API response, proving source bytes were not altered
• Enriched — optional LLM-generated context.summary in the declared language
• Linked — every entity is a dereferenceable HTTP URI returning JSON-LD (NEW in v0.2.0)

2. Linked Data principles

CDS v0.2.0 is designed to satisfy the four Linked Data rules articulated by Tim Berners-Lee.

Rule 1: Use URIs as names for things. CDS names every entity with a URI. Events are identified by https://signed-data.org/events/{uuid}, content types by https://signed-data.org/vocab/{domain}/{schema}, and sources by https://signed-data.org/sources/{source-id}.

Rule 2: Use HTTP URIs so people can look up those names. All CDS URIs use the https://signed-data.org base. They are not opaque identifiers — they are HTTPS URIs that resolve to real endpoints on the web.

Rule 3: When someone looks up a URI, provide useful information using standards. Every CDS URI returns a JSON-LD document. Source URIs return source metadata; content type URIs return schema definitions; the vocabulary root returns the full ontology.

Rule 4: Include links to other URIs. CDS events contain links to other CDS resources — events to sources to domains to vocabulary, forming a navigable graph.

See Linked Data for the deep-dive.

3. 5-star open data rating

CDS v0.2.0 achieves the maximum 5-star open data rating:

★       Available online under an open license              ✓ (v0.1.0)
★★      Structured data (not a scanned image)               ✓ (v0.1.0)
★★★     Non-proprietary format (JSON)                       ✓ (v0.1.0)
★★★★    Use open W3C standards (JSON-LD)                    ✓ NEW in v0.2.0
★★★★★   Link to other people's data (URI links)             ✓ NEW in v0.2.0

4. Event envelope

Every CDS v0.2.0 event is a valid JSON-LD document:

{
  "@context": "https://signed-data.org/contexts/cds/v1.jsonld",
  "@type":    "https://signed-data.org/vocab/CuratedDataEvent",
  "@id":      "https://signed-data.org/events/a3f8c2d1-4e2b-4f8a-9c1d-2e3f4a5b6c7d",

  "spec_version": "0.2.0",
  "id":           "a3f8c2d1-4e2b-4f8a-9c1d-2e3f4a5b6c7d",

  "content_type": "https://signed-data.org/vocab/lottery-brazil/mega-sena-result",

  "source": {
    "@id":         "https://signed-data.org/sources/caixa.gov.br.loterias.v1",
    "fingerprint": "sha256:b310..."
  },

  "occurred_at":  "2026-03-29T00:00:00Z",
  "ingested_at":  "2026-03-29T00:00:04Z",
  "lang":         "pt-BR",
  "payload":      { },

  "context": {
    "summary":      "Mega Sena concurso 2800...",
    "model":        "rule-based-v1",
    "generated_at": "2026-03-29T00:00:05Z"
  },

  "integrity": {
    "hash":      "sha256:a1b2c3...",
    "signature": "MX6rj3...",
    "signed_by": "https://signed-data.org"
  }
}

The @context, @type, and @id fields MUST appear first in serialised output. The id field (plain UUID) is retained alongside @id (full URI) for convenience.

5. Field reference

Field	v0.1.0	v0.2.0	Change
@context	absent	"https://signed-data.org/contexts/cds/v1.jsonld"	new
@type	absent	"https://signed-data.org/vocab/CuratedDataEvent"	new
@id	absent	"https://signed-data.org/events/{uuid}"	new
spec_version	"0.1.0"	"0.2.0"	bumped
content_type	CDSContentType object	URI string	breaking
source.id	opaque string	absent	removed
source.@id	absent	HTTP URI	replaces source.id
integrity.signed_by	"signed-data.org"	"https://signed-data.org"	breaking

6. URI scheme

All CDS URIs share the base https://signed-data.org.

Resource	Pattern	Example
Event	/events/{uuid}	https://signed-data.org/events/a3f8c2d1-...
Content type	/vocab/{domain-slug}/{schema-slug}	.../vocab/lottery-brazil/mega-sena-result
Source	/sources/{source-id}	.../sources/caixa.gov.br.loterias.v1
Vocabulary root	/vocab/	https://signed-data.org/vocab/
JSON-LD context	/contexts/cds/v1.jsonld	.../contexts/cds/v1.jsonld
Public key	/.well-known/cds-public-key.pem	.../cds-public-key.pem

Domain slugs: dots in the domain name become hyphens. lottery.brazil → lottery-brazil. sports.football → sports-football.

Schema slugs: dots in the schema name become hyphens. mega-sena.result → mega-sena-result.

Source IDs: dots are kept as-is. caixa.gov.br.loterias.v1 stays caixa.gov.br.loterias.v1.

7. Content type URIs

https://signed-data.org/vocab/{domain-slug}/{schema-slug}

The builder function signature is content_type_uri(domain, schema_name) — both arguments use the original dot notation.

Registered content types

lottery-brazil

Constant	URI
LOTTERY_MEGA_SENA	.../vocab/lottery-brazil/mega-sena-result
LOTTERY_LOTOFACIL	.../vocab/lottery-brazil/lotofacil-result
LOTTERY_QUINA	.../vocab/lottery-brazil/quina-result
LOTTERY_LOTOMANIA	.../vocab/lottery-brazil/lotomania-result
LOTTERY_DUPLA_SENA	.../vocab/lottery-brazil/dupla-sena-result

sports-football

Constant	URI
FOOTBALL_MATCH_RESULT	.../vocab/sports-football/match-result
FOOTBALL_MATCH_LIVE	.../vocab/sports-football/match-live
FOOTBALL_STANDINGS	.../vocab/sports-football/standings-update

weather

Constant	URI
WEATHER_CURRENT	.../vocab/weather/forecast-current
WEATHER_DAILY	.../vocab/weather/forecast-daily
WEATHER_ALERT	.../vocab/weather/alert-severe

finance / finance-brazil

Constant	URI
FINANCE_STOCK	.../vocab/finance/quote-stock
FINANCE_CRYPTO	.../vocab/finance/quote-crypto
FINANCE_FOREX	.../vocab/finance/quote-forex
FINANCE_INDEX	.../vocab/finance/index-update

See the finance.brazil domain spec for the full Brazilian finance content types.

8. Source registry

A source represents a verified upstream data provider. Each source is identified by a URI of the form https://signed-data.org/sources/{source-id}.

Source document

Dereferencing a source URI returns a JSON-LD document:

{
  "@context": "https://signed-data.org/contexts/cds/v1.jsonld",
  "@type":    "https://signed-data.org/vocab/DataSource",
  "@id":      "https://signed-data.org/sources/caixa.gov.br.loterias.v1",
  "name":     "Caixa Econômica Federal — Loterias",
  "domain":   "caixa.gov.br",
  "version":  "v1",
  "upstream": "https://servicebus2.caixa.gov.br/portaldeloterias/api/megasena",
  "content_types": [
    "https://signed-data.org/vocab/lottery-brazil/mega-sena-result",
    "https://signed-data.org/vocab/lottery-brazil/lotofacil-result"
  ]
}

Registered sources

Constant	Source URI
CAIXA_LOTERIAS	.../sources/caixa.gov.br.loterias.v1
API_FOOTBALL	.../sources/api-football.com.v3
OPEN_METEO	.../sources/open-meteo.com.v1
BRAPI	.../sources/brapi.dev.v1
BCB_SGS	.../sources/api.bcb.gov.br.v1
CONAB	.../sources/conab.gov.br.v1
WORLD_BANK	.../sources/api.worldbank.org.v2
BRASILAPI	.../sources/brasilapi.com.br.v1

9. Vocabulary

The CDS ontology is published at https://signed-data.org/vocab/. It defines the classes and properties used in CDS events.

Classes

Class	URI	Description
CuratedDataEvent	.../vocab/CuratedDataEvent	The top-level event envelope
IntegrityMeta	.../vocab/IntegrityMeta	Cryptographic proof attached to an event
DataSource	.../vocab/DataSource	A registered upstream data provider

Properties

Property	URI	Domain	Range
specVersion	.../vocab/specVersion	CuratedDataEvent	xsd:string
contentType	.../vocab/contentType	CuratedDataEvent	URI
source	.../vocab/source	CuratedDataEvent	DataSource
occurredAt	.../vocab/occurredAt	CuratedDataEvent	xsd:dateTime
ingestedAt	.../vocab/ingestedAt	CuratedDataEvent	xsd:dateTime
lang	.../vocab/lang	CuratedDataEvent	xsd:string
payload	.../vocab/payload	CuratedDataEvent	JSON object
context	.../vocab/context	CuratedDataEvent	JSON object
hash	.../vocab/hash	IntegrityMeta	xsd:string
signature	.../vocab/signature	IntegrityMeta	xsd:string
signedBy	.../vocab/signedBy	IntegrityMeta	URI
fingerprint	.../vocab/fingerprint	DataSource	xsd:string

10. Signing algorithm

1. Serialise to canonical JSON — sort_keys=True, UTF-8, excluding integrity and ingested_at
2. hash = "sha256:" + SHA256(canonical_bytes).hexdigest()
3. Sign canonical_bytes with RSA-PSS (SHA-256 digest, MGF1, max salt length)
4. Encode signature as base64
5. Attach IntegrityMeta { hash, signature, signed_by }

Verification: re-serialise canonical bytes, check hash, verify RSA-PSS signature with the issuer public key published at https://signed-data.org/.well-known/cds-public-key.pem.

v0.2.0 canonical bytes

The canonical bytes now include the @context, @type, and @id fields (which did not exist in v0.1.0). The exclusion set remains integrity and ingested_at only. A v0.1.0 event cannot be verified by a v0.2.0 verifier, and vice versa, because the canonical bytes differ.

For full details, see Signing.

11. JSON-LD compatibility

The @context at https://signed-data.org/contexts/cds/v1.jsonld maps the snake_case JSON keys used by CDS to RDF predicates:

{
  "@context": {
    "@vocab":       "https://signed-data.org/vocab/",
    "spec_version": "specVersion",
    "content_type": { "@id": "contentType", "@type": "@id" },
    "source":       { "@id": "source",      "@type": "@id" },
    "occurred_at":  "occurredAt",
    "ingested_at":  "ingestedAt",
    "signed_by":    { "@id": "signedBy",    "@type": "@id" },
    "fingerprint":  "fingerprint",
    "lang":         "lang",
    "payload":      "payload",
    "context":      "context",
    "hash":         "hash",
    "signature":    "signature",
    "integrity":    "integrity"
  }
}

This achieves two goals simultaneously:

1. Plain JSON consumers are unaffected. A consumer that ignores @context, @type, and @id reads all other fields exactly as before. No JSON-LD library required.
2. RDF consumers get valid triples. A JSON-LD processor expands any CDS event into an RDF graph. Fields typed as @id produce proper RDF links.

12. Breaking changes from v0.1.0

Change	v0.1.0	v0.2.0	Impact
content_type field type	CDSContentType object	URI string	All code that reads content_type.domain must change
source.id renamed to source.@id	opaque string	HTTP URI	JSON key change; use SDK helpers
integrity.signed_by	"signed-data.org"	"https://signed-data.org"	String comparison will break
Canonical bytes	Exclude integrity, ingested_at	Same exclusions, but include @context, @type, @id	v0.1.0 signatures cannot be verified by v0.2.0 verifiers
New required fields	N/A	@context, @type, @id	Producers must include these fields

For migration code examples, see Migration v0.1 → v0.2.