Skip to main content

Specs & Standards

FleetForge aligns its receipts, telemetry, and provenance story to three public specs so reviewers can trace every promise back to a canonical citation.

OpenTelemetry GenAI​

We emit budgets, token usage, and replay spans using the names defined in the OpenTelemetry GenAI semantic conventions and metrics appendix. That means gen_ai.client.token.usage and gen_ai.cost.usd in /demo, /product, and the CLI line up with the same fields collectors and observability teams already expect.

C2PA Content Credentials​

Every artifact (hello-fleet outputs, manifests, optional attachments) ships with a C2PA manifest so downstream tools can verify provenance without FleetForge. The spec lives at the C2PA Content Credentials site; we follow the JSON-LD profile for AIBOM + attestation bundles, with optional SCITT anchors pointing to the same manifest IDs.

SCITT transparency​

SCITT provides the transparency feed backing optional external verification. The reference implementation and profile live under the IETF SCITT working group. When SCITT is enabled, FleetForge publishes the same manifest_digest + policy verdict metadata so third parties can resolve claims independently.

Use this page whenever you need to cite the underlying standards in reviews, incident reports, or acceptance docs—each section mirrors the references wired into the demo proof rail and landing-page copy.