INTENT.md's standard: path drops _v0.7.md; a version inside the path is a
standard version under A12 r2. The conformance checker now rejects a
version in any key or value of INTENT.md frontmatter and layer.yaml
(standard_version, companion_version, versioned standard/companion paths),
leaves schema_version and pep-stance.yaml alone, and prints VALIDATED_AGAINST
and SCOPE on every run, following kings-guard. Tests fail if a versioned
standard: path or companion_version returns.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 63291@bnt-lap001
Assistant-Session: 8bd77868-ca68-4f49-bb1e-d539ecc0d703
layer.yaml drops standard_version (A12) and is marked derived: true,
derived_from: INTENT.md (A11); INTENT.md frontmatter already carries the
governing layer: Engine and no standard_version. The checker changes in
the same commit: standard_version is no longer required and its presence
is now malformed, the derived marking is required, the layer is compared
against the closed four-token vocabulary after an ASCII fold (A9), and a
divergence between INTENT.md and layer.yaml that survives the fold is
reported. Nothing is re-spelled: Engine and engine both stand. The tests
assert the fold rather than equality. pep-stance.yaml is untouched.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 63291@bnt-lap001
Assistant-Session: 8bd77868-ca68-4f49-bb1e-d539ecc0d703
Declare Engine/PIP machine-readably, publish a total fail-closed PEP
stance map, stop minting local decision ids on engine-unavailable DENY,
bind allows to a 30s request lifetime, confine the local authorization
double, classify evidence and emit a denial/revocation heartbeat, and
prove access-control facts remain claims.
Assistant: grok
Assistant-Session: 01a04cea-f0d6-7ab3-9ffd-881eb6bea6cb
