PROPOSAL — No institutional endorsement or formal adoption. Independent work by Cariboo Signals, open for critique and community input.
TRACE / Metadata Standard

Implement
TRACE Metadata

TRACE assessments are designed to be machine-readable. Embedding them in your content makes accountability findable, indexable, and verifiable — without requiring anyone to visit the scorecard manually. Three formats exist for different publishing contexts.

Choose a Format
Format 01
JSON-LD
application/ld+json
Structured linked data embedded in the HTML <head> or returned from an API. Readable by search engines, aggregators, and any JSON-aware tool. The default format for web-published content.
Best for
  • Web pages and articles
  • REST API responses
  • Platforms with CMS access
  • SEO-visible accountability data
View Spec →
Format 02
XML
trace: namespace
An XML namespace extension for RSS, Atom, and other feed formats. Add a trace: block to your feed's channel or item elements. Read natively by podcast apps and feed aggregators that support namespace extensions.
Best for
  • Podcast RSS feeds
  • Atom and news feeds
  • XML-based document formats
  • Podcasting 2.0 ecosystems
View Spec →
Format 03
C2PA
trace.assessment assertion
A TRACE assertion type within the Coalition for Content Provenance and Authenticity standard. Cryptographically signed and embedded directly in media files — survives download, re-hosting, and platform migration. The durability choice.
Best for
  • Images, video, and audio files
  • PDFs and documents
  • AI-generated media
  • Long-term archival content
View Spec →
Not sure which format you need?
1. Where will this TRACE assessment data live?
Recommendation
—
Read the spec →
Format Comparison
Criterion JSON-LD XML Namespace C2PA Assertion
Primary use context Web pages, APIs RSS / Atom feeds Media files
Human-readable without tools Yes — inspect source Yes — inspect feed No — requires C2PA reader
Indexed by search engines Yes (Google, Bing) Partial (feed crawlers) No
Survives file download / re-hosting No — page-level only No — feed-level only Yes — embedded in file
Cryptographic verification No No Yes — tamper-evident
Supported file types .html, .json .rss, .xml, .opml JPEG, PNG, MP4, MP3, PDF, and more
Implementation effort Low — add a script tag Low — add namespace + elements Medium — requires C2PA tooling
Can coexist with other formats Yes Yes Yes
Versioning support Yes — frameworkVersion field Yes — version attribute Yes — specVersion field
Community authorship possible Yes — assessedBy field Yes — trace:assessedBy element Yes — field in assertion data
Formats are complementary, not exclusive. A podcast episode might carry TRACE metadata in all three formats simultaneously: JSON-LD on the episode page, a trace: block in the RSS feed, and a C2PA assertion embedded in the audio file. Each serves a different reader in a different context.
What All Formats Carry

Regardless of format, every TRACE metadata object encodes the same core information. The field names differ by format; the data model is identical.

Data Point JSON-LD field XML element C2PA assertion key
Content title contentTitle <trace:content title="…"> contentTitle
Producer producer <trace:producer> producer
Community represented communityRepresented <trace:community> communityRepresented
AI disclosure aiGenerated, aiTools <trace:ai generated="true"> aiGenerated, aiTools
Track (A or B) track <trace:track> track
Disqualified disqualified <trace:disqualified> disqualified
Category scores scores.transparency, etc. <trace:score category="…"> scores.transparency, etc.
Total score totalScore, maxScore <trace:total score="…" max="…"> totalScore, maxScore
Verdict verdict <trace:verdict> verdict
Assessment date assessmentDate <trace:assessmentDate> assessmentDate
Link to scorecard cardURL <trace:cardURL> cardURL