Skip to content

P5_A0_STANDARDS_COMPOSITION_DECISION.md

Latest commit

 

History

History
73 lines (50 loc) · 3.76 KB

P5_A0_STANDARDS_COMPOSITION_DECISION.md

File metadata and controls

73 lines (50 loc) · 3.76 KB
Raw

P5-A0 — Standards Composition Decision

Status: SCIENTIFICALLY CLOSED — PASS BOUNDED / RESEARCH-ONLY

Decision:

P5_EXISTING_STANDARDS_COMPOSITION_SUFFICIENT_BOUNDED

Accepted source:

  • commit: 9ff34d9baaa01a90cf6d2cb95e1930dc4c011e11
  • workflow run: 36743125582
  • evidence artifact: 11110384934
  • artifact SHA-256: b769b01ccd6051d23fbf998d37c30d9d7073f39505301f4cec5fabbdfada3595

What was established

For the bounded P5-A0 fixture, existing standard predicates are sufficient to represent the required ExecSurface verification graph without inventing a new in-toto predicate:

  • in-toto Statement v1;
  • Runtime Trace v0.1 for the bounded runtime observation representation;
  • SCAI v0.3 for detailed ExecSurface-specific verification conditions and evidence binding;
  • SVR v0.2 for the concise verification result over the same subject;
  • optional exact reference to SLSA provenance/v1 by predicate type, statement digest and subject digest.

This establishes representational sufficiency only. It does not establish generic-consumer understanding of ExecSurface authority semantics, signature trust, SLSA level, public-release readiness or multi-backend equivalence.

Frozen A0 gate result

The preregistered corpus remained exactly 18 tests. The accepted run passed:

  • P5-A0: 18/18;
  • P4-C import corpus: 14/14;
  • P4 cross-proposition falsification: 12/12;
  • P4-B1 live ptrace authority: 10/10;
  • P4-B2 live ptrace authority: 10/10;
  • P4-B3 live ptrace authority: 10/10;
  • Semantics v3 reproof: 7/7;
  • public M11 fail-closed reproof: 6/6;
  • rustfmt and clippy -D warnings: PASS;
  • immutable alpha.4 / stable v0.1 boundaries: PASS;
  • pinned in-toto standards snapshot identity: PASS.

The strengthened test #16 explicitly exercises post-build substitution of the embedded SLSA provenance digest and predicate type and requires verification to fail closed. The test count, threshold and semantic success criteria were not changed.

Retained negative / pre-close evidence

No failed or weaker run is erased or reclassified as a scientific PASS.

  • run 36741656523: static rustfmt failure before scientific corpus; artifact 11109703891, SHA-256 d41c33fec5f8a0c12cd6b8f2bfa764772ffb34afd1d73e6dab8264931bb01090;
  • run 36742010054: remaining test-file rustfmt failure before corpus; artifact 11109699513, SHA-256 3ae560b05261f4f294b4a27243cf5ecf3bb2a7a7f5681adf0c507331295e2dcc;
  • run 36742159476: compile/type defect in bounded subject comparison before corpus;
  • run 36742784570: full green predecessor run, but retained as PRE-CLOSURE / TEST-HARNESS-ADEQUACY-DEFECT because the original test #16 proved identity sensitivity rather than post-build fail-closed substitution. Artifact 11111355567, SHA-256 5b0fb24720d68b918ddeeb1269e356ce4299c984b9c8c2bdd593eeac5b9aad63.

The accepted run is 36743125582, after the adequacy correction.

Scientific boundaries

This decision does not authorize:

  • a new ExecSurface predicate;
  • signing or attestation authenticity as semantic authority;
  • treating a SLSA provenance reference as runtime behavioral truth;
  • assigning a SLSA level;
  • promotion into public crates, public observer behavior, v2 semantics, v0.1.0-alpha.4, or stable @v0.1;
  • interpreting backend/profile names as authority;
  • treating incomplete, ambiguous, lost or unsupported evidence as PASS.

The research-only implementation remains isolated under experiments/p5-attestation-provenance/.

Next gate

Proceed to P5-A1 — Runtime Trace Mapping Gate under the already frozen P5 protocol. A1 must prove the mapping itself, including fail-closed cross-binding of monitor identity, exact command/job context, monitor-log mutation, observer loss/truncation and replay. A0 is not reopened unless a new counterexample invalidates this decision.