Skip to content
Open Intent Schema

Appendix C: Open questions

These questions are open for v0.1 and are the main topics for the first design partner rounds.

  • Non-functional requirements: is kind: non-functional enough, or do NFRs need structured, measurable thresholds such as a latency budget?
  • One evidence per criterion: v0.1 binds each evidence record to one criterion, so a test run covering three criteria produces three records. Is that practical?
  • Actor granularity: v0.1 treats the model as an attribute of the actor and requires a new actor version when it changes. Should the model be an actor of its own?
  • Human approval: should the core require a human actor for intent-approval and release decisions, or only regulated profiles?
  • Proof of derivation: derivedFrom: intent relies on a complete context. How can producers show completeness beyond declaring it?
  • Evidence across versions: is the digest-based rule in section 6.3 workable in practice?
  • Git digests: gitCommit is SHA-1 in most repositories. Should a sha256 value be required as well?
  • Autonomy levels: do the four levels belong in the core or in a profile?
  • Identifiers: project domain, passport predicate type URI, event type prefix, IANA media type registration and a possible URN namespace.
  • Name: “Open Intent Schema” is a working title and needs a trademark check.
  • Foundation: which neutral home fits best, for example the Linux Foundation, CNCF or OpenSSF?
  • Runtime attestation: partly addressed in draft 2 by R12 and the actor identity. Still open: how a runtime proves that it recorded context faithfully, for example through sandbox attestation or signed session logs.
  • IPR policy: which patent covenant should cover the specification text and the schema?
  • Reference method: Appendix D now defines an informative method. Should it become normative in 1.0?