secrets-engine's first live request rejected our allow: binding.
request_digest is computed over material they never sent, because we
enrich subject and resource from the registry before hashing. Answering
that meant reading the enrichment path, which had a worse defect in it.
Enrichment was additive-if-absent — addAttribute wrote a registry value
only where the request had no value for that key. So where a caller
supplied a key, the caller's value won and the registry's never applied.
Every registry ceiling and allowlist was advisory. Verified against the
shipped ops-warden package, each one added key on an otherwise-denied
request:
max_ttl_hours: 99 registry says 8 -> allowed a 12h certificate
allowed_principals registry allowlist -> disallowed_principal bypassed
allowed_subjects registry allowlist -> unknown_subject bypassed
The third is the one to read twice: a subject the registry does not know
authorized itself by naming itself in the allowlist it was being checked
against.
Not remotely reachable today — the PEP builds the CheckRequest,
ops-warden sends no resource.attributes, and enforce admits one identity.
It is a defence-in-depth failure: any path that lets attacker-influenced
data into a CheckRequest field became a full policy bypass rather than a
bounded input problem. Callers sending resource.attributes is not
hypothetical; secrets-engine does it on every request.
Registry facts now win, and diagnostics.registry_overrode names every
displaced key, because a registry that silently discards a contradicting
claim hides that a caller asserted authority it did not have.
subject.type is carved out, and the reason is a finding of its own.
Making the registry win there denied every secrets-engine allow: the
registry's type is CARING vocabulary (Human, Agent, Automation, Service)
and the request's is the protected system's actor vocabulary (service,
adm, agt, atm). Two fields sharing a name; substituting one for the other
is translation rather than identity, which GH-DEC-2026-008 ruled against.
Note what surfaced it — the registry's type had been dead data since the
field existed, because the caller's value always won.
Also publishes binding.submitted_request_digest, over the request exactly
as sent. request_digest was published as the consumer replay test and
cannot be one. Nothing is lost hashing the pre-enrichment form:
enrichment is a function of the request and the snapshot, and
registry_snapshot_digest already pins the snapshot.
Existing pins do not move. All three replay fixtures' request_digest and
approval_binding_digest values are byte-identical — those requests
contradict no registry fact. A field to add, not a value to correct.
Regression tests verified failing against the old behaviour before being
kept. FLEX-DEC-2026-012; FLEX-WP-0025 carries the residual, that a policy
still cannot tell a fact from an assertion.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014aQMM1dPXaPiXVn6DwwtLd
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014aQMM1dPXaPiXVn6DwwtLd
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
Updated by fix-consistency on 2026-09-07:
- update .custodian-brief.md for flex-auth
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
gate-house circulated security-layer-model v0.8 for assent. Assent given
with four findings; F1 is fail-open and blocking for calling the section
6.4 PEP obligations complete, though not for anyone's adoption — the
condition already exists under v0.7 and is not created by this version.
F1. "A decision from access-engine" is not a checkable predicate. 6.4
obligation 1 requires a PEP to hold one; obligation 2 supplies a test it
is emphatic is mechanical and not a judgement call. That test establishes
which request a decision is for and nothing about who issued it, because
every input to every digest a PEP can check is either sent by the caller
or published. A responder knowing the package id and version reproduces
all three exactly.
What makes it sharp is internal: 9.4 requires the approval object to
carry authenticated entries, and 6.4 obligation 5 then requires each
artifact be validated against its owning layer, naming the approval-claim
and DecisionEnvelope pair as the live instance. So the obligation is
unsatisfiable for half the pair it was written about — a PEP can validate
the approval's authenticity and cannot validate the decision's. The
asymmetry is argued nowhere.
Section 16 already carries this observation one layer up, about the
standard's own publication path having no digest or freeze discipline.
Applied to the artifact the standard regulates, it is F1. The gap was
visible from inside and got recorded against the wrong artifact.
F2 answers the absent-vs-unknown question gate-house flagged as ours.
Runtime behaviour must be identical — absent fails closed too, and for a
stronger reason, since it is the branch reached by discovering your model
of your own scope axis was wrong. But omitting the distinction makes
totality unfalsifiable: a map with an unknown catch-all satisfies 6.4
obligation 3 vacuously, and its drift test passes by exercising the
catch-all rather than the axis. Same defect we published as
FLEX-DEC-2026-008 — coverage counted rather than executed, in a stance
map instead of a fixture suite.
F3. "The published map MUST equal the shipped behaviour, and that
equality SHOULD be asserted by a test", in a paragraph arguing drift is
worse than no publication. The strongest obligation in 6.4 has the
weakest verification, and 13.1 shows four of five rows already test it.
F4. 13.1 marks ops-warden's unknown cell non-conformant and states
ops-mason's map as simply not published, which is a plainer violation of
the same obligation, unmarked in the same table.
Also answered their direct question: no, they did not over-read our
divergence report. Added that the ruling costs us nothing, since we are
not PEP-shaped, so our concurrence is weak evidence and ops-warden's
assent is worth more here.
FLEX-WP-0023-T05 and FLEX-WP-0024-T04 close. Notes T01-T03 landed from
Glas meanwhile with real receipts including an actually-expired token,
and that enforce authenticates the caller while leaving F1 untouched —
as their own note says.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014aQMM1dPXaPiXVn6DwwtLd
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014aQMM1dPXaPiXVn6DwwtLd
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
Updated by fix-consistency on 2026-09-06:
- update .custodian-brief.md for flex-auth
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
Updated by fix-consistency on 2026-09-06:
- workplan status: ready → active
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
secrets-engine probed the Service DNS name handed over in FLEX-WP-0021-T05
and found it resolves, from the workstation, to an unrelated public host.
Reproduced here: search ad.binect.de answers wildcard, so
flex-auth-secrets-engine.flex-auth.svc.cluster.local and
this-service-does-not-exist.flex-auth.svc.cluster.local both resolve to
80.158.43.29, while the trailing-dot FQDN correctly fails. A bare Service
name in a handover is not merely unreachable from there, it is a live
misdirection, and the handover was ours.
Had a deployment pointed at it, the CheckRequest body would have gone to
that host: subject, tenant, lane and resource ids, stage, field names,
purpose, plus the caller's bearer token.
Trailing-dot FQDN and "in-cluster only" now replace the bare name in the
example README, SCOPE.md, and the T05 note.
Their real question was how the response channel is authenticated, and
they declined to answer it locally because choosing a transport control
for our service is not a consumer's call. Right boundary, so the answer
is recorded here as FLEX-DEC-2026-010: it is not authenticated. Pins
serve plain HTTP, the envelope carries no signature, and a responder that
knows the package id and version can return a well-formed allow that
passes every check a consumer performs.
The part worth stating in the contract is that the digests do not help
and look like they do. Every input to request_digest,
policy_package_digest and registry_snapshot_digest is either sent by the
caller or published in this repo, so a forger reproduces all three
exactly. They establish integrity of the binding, never authenticity of
the source — and publishing more digests makes a forged envelope look
more authenticated, not less.
For secrets-engine specifically: fail-closed protects against a PDP that
is absent, not against one that lies. An unreachable PDP denies; a lying
PDP allows.
Third instance of one seam in three decisions. 008: a tenant carried
into the digest and never compared — visible, not enforced. 009: a caller
authenticated and never recorded — enforced, not visible. 010: a record
verifiable and unauthentic — checkable, but not evidence.
One nuance that changes the operator recommendation: kubectl port-forward
does authenticate the responder, transitively — no DNS name, one named
pod, API-server TLS. That is the exact reverse of the caller direction,
where it bypasses the NetworkPolicy. Independent properties pointing
opposite ways, so neither can be summarised as "the network protects it".
FLEX-WP-0024 carries signing; key custody routes through warden/OpenBao
rather than minting a key here.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014aQMM1dPXaPiXVn6DwwtLd
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014aQMM1dPXaPiXVn6DwwtLd
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
Updated by fix-consistency on 2026-09-06:
- update .custodian-brief.md for flex-auth
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
glas-harness asked for a supported owner access path with authenticated
caller binding and lifetime, ruling out Service DNS and a permanent
operator token. Both refusals are correct and the answer needs no new
mechanism: a TokenRequest token is short-lived, audience-scoped, and
bound by exact sub to the ServiceAccount the deployed pin already names.
Two findings came out of designing it.
First, for an operator caller the pin's NetworkPolicy is not a partial
control, it is silent: kubectl port-forward is proxied to the pod's own
loopback and no policy selector is consulted. So under warn the operator
path is unauthenticated and unfiltered, and the only reason it is not
reachable is that nobody has forwarded the port. warn was safe for
ops-warden because a workload pin still admitted one pod; here the warn
window protects nothing while it runs. enforce is the deliverable.
Second, the ServiceAccount named by the deployed binding does not exist.
Namespace secrets-engine holds only default, so enforce today would deny
every request rather than authenticate one.
Third, and this is the contract one: the decision record has no caller
field. provenance carries evaluator, mode, policy and registry digests
and decision time; binding carries the normalized request. So the four
negative tests can all pass and no artifact retains that they passed for
the request that mattered. Same seam as FLEX-DEC-2026-008 one layer up —
there a tenant was carried into the digest and never compared, visible
but not enforced; here a caller is authenticated and never recorded,
enforced but not visible.
provenance.caller, not binding.caller: the same request from a different
authenticated caller must decide identically, so the caller is not
replay identity and must not move request_digest. FLEX-DEC-2026-009.
Live receipts are not included. kubectl create token is credential
minting and was refused in this session, correctly; the commands are
exact and the expectations read out of internal/callerauth/auth.go, but
nothing is claimed as verified that was not run.
Also records the operator's tenant:platform decision
(5ed3fb35-eca9-413a-82b9-95171ba85bf6) and verifies that v2 already
enforces exactly it — exact string equality, no alias, no normalisation.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014aQMM1dPXaPiXVn6DwwtLd
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014aQMM1dPXaPiXVn6DwwtLd
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
Updated by fix-consistency on 2026-09-06:
- update .custodian-brief.md for flex-auth
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
secrets-engine.catalog-lane.lifecycle v1 contained no reference to
input.tenant — not in well_formed, not in the denial ladder, not in a
test. A rotate on lane:glas-primary under tenant:coulomb returned allow
against the deployed package (decision:066e629bbf0c0924).
Found while answering glas-harness's tenant-alignment request, which had
asked for wrong-tenant denial evidence. There was none to return.
Three covers failed the same way: every one of the 29 fixtures carried
tenant:platform, so the suite could not report on the field; T02's own
gate named "wrong-tenant deny" and was recorded done unmet; and the
engine hashes tenant into request_digest but never compares it. Four
other published packages carry the branch — this one was the outlier.
v2 adds wrong_tenant above wrong_system, three Rego tests and three
fixtures (28/28, 32/32). The absent-tenant test caught a second defect
in the first draft: a bare input.tenant != comparison is undefined on a
missing key, so the branch dropped and the ladder reported the wrong
rung. request_tenant := object.get(input, "tenant", "") fixes it.
v2 supersedes rather than amends v1 because the defect failed open: a
consumer pinned to _VERSION=v1 would keep receiving allows with no
signal the rule beneath the version string had changed. The earlier
dual-control correction stayed at v1 because it denied everything.
Replay envelopes regenerated at v2; both request_digest values are
byte-identical, so secrets-engine's digest join needs no re-pinning.
The sweep this prompted found tenant-engine unscoped on tenant as well —
deployed, and verified allowing tenant:coulomb. Not the same fix: its
request tenant names the target rather than the caller, so a constant
would break it. Recorded and carried by FLEX-WP-0022 rather than patched
unilaterally.
FLEX-WP-0021 closes at T05; the pin still serves v1 until a redeploy.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_014aQMM1dPXaPiXVn6DwwtLd
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715613@bnt-lap001
Assistant-Session: fabd95c1-4c9e-4080-8849-8707ae025f80
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
Updated by fix-consistency on 2026-09-06:
- update .custodian-brief.md for flex-auth
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
secrets-engine confirmed T03, and re-verifying against the regenerated
destroy fixture found something neither repository can fix alone: a
pdp_digest recorded at issue time can never equal the request_digest of a
request that carries the claim in its context, because the claim is part
of the context that is hashed. Embedding the claim changes the very
digest the claim would need to name.
Not fixture staleness. It holds for every dual-control request whose
claim travels in context -- the shape GH-DEC-2026-008 had just ruled
mandatory. Left unresolved that ruling was unimplementable for exactly
the case it was written for, and destroy would have been permanently
un-allowable in production, failing closed forever on a check that could
never pass.
flex-auth owns the canonical request digest, so the fix is ours.
binding.approval_binding_digest is the same material with
context.approval removed, emitted only when a claim was carried. An
approval issued against a claim-free Check records that Check's
request_digest; the claim-bearing request reproduces it here.
DELIBERATELY ADDITIVE, and the reason matters. The tempting fix is to
drop context.approval from request_digest entirely. That is wrong:
request_digest is the replay identity, and two requests differing only in
which approval was presented must not share one, because their decisions
differ -- one allows, the other denies dual_control_required. Collapsing
them would let an allow obtained with a valid claim be replayed against a
request carrying none. So request_digest still covers the claim and still
moves; approval_binding_digest deliberately does not, and is documented
as not a replay identity. The tests assert the two functions DISAGREE on
a claim-bearing request, which is approval-engine's formulation of how to
defend a distinction that looks like duplication.
The fixture now demonstrates the property rather than asserting it: its
claim's pdp_digest equals the envelope's approval_binding_digest with
pdp_path true, and changing the claim's contents moved request_digest
while leaving approval_binding_digest untouched. Two files a consumer can
diff.
Also picked up approval-engine's new required binding.pdp_path via the
cross-repo schema test added yesterday -- which is the test doing exactly
what it was built for, one day later.
T03 is done. secrets-engine's own digest-material defect, which our two
real envelopes caught, is recorded in the workplan.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
approval-engine suggested a conformance check after the examples-contradict-
prose class hit a third repository -- theirs. Fifteen lines, they said, and
it would have caught our caring fixture and our provenance omission. Worth
stealing, so we stole it.
internal/schemaguard validates published examples against published schemas.
It implements only the JSON Schema subset these schemas use, and the property
that makes it trustworthy is that an unrecognised keyword FAILS rather than
skips: a validator that silently approves what it does not understand invites
reliance it cannot support. It found three things on first run.
ONE, AND THE LARGEST: check_request.schema.json pointed subject.type at
CARING's subject_type enum (Human, Service, ...), and no consumer sends that
vocabulary. user-engine sends human, tenant-engine and secrets-engine send
service, and ops-warden sends adm/agt/atm -- an actor-type vocabulary CARING
does not model at all. Our published schema declared three live integrations
non-conformant. A rule that outlaws shipped correct behaviour is the rule
that is wrong, so the $ref is replaced with an opaque non-empty string and a
description saying why. CARING's enum remains correct where it belongs: the
registry's subject_manifest.yaml, where Service is right.
TWO: policy_package_note, which this session added to the caring example's
decision provenance, is undeclared under additionalProperties:false. Our own
annotation broke the conformance it was annotating. Moved to the envelope's
outer provenance.
THREE: the secrets-engine fixtures carried partial approval-claims, missing
binding, freshness and validity. A partial claim in a fixture is how a
consumer learns the wrong shape -- the same mechanism that produced the
destroy defect. They are now complete and valid against approval-engine's
schema, including the now-required binding.pdp_digest, and a test validates
them against that schema when the sibling repo is present.
The destroy replay fixture is regenerated accordingly and the replay README's
pinned digests updated, since a stale digest table is the same defect wearing
a different hat.
Closed the binding-mapping open item. approval-engine declined to publish a
vocabulary mapping and their reasoning is better than the request: a PIP
asserting secrets.kv.destroy MEANS destroy would author semantics over two
vocabularies it owns neither of, and a wrong mapping silently accepts a claim
approved for a different action. pdp_digest is the mapping precisely because
it does not translate. It is now always present and nullable, so the destroy
gate is pdp_digest non-null and equal, enforced at the PEP.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
TWO FINDINGS, BOTH FROM READING OTHER REPOSITORIES RATHER THAN OURS.
The §13.1 stance register has a second row. secrets-engine published
pep-stance.yaml -- total over catalog stage plus unknown, runtime-read
and pinned to SHIPPED_STANCE by test. SCOPE.md claimed ops-warden's was
the estate's only published map and that the register's single row was
itself the finding; that is no longer true and is corrected.
docs/stance-register-review.md is the first exercise of the
aggregate-divergence capability flex-auth claimed on 2026-08-29 and then
recorded as unexercised because one row cannot diverge from anything.
Three findings:
The two maps take opposite stances on unknown -- ops-warden fail_open by
versioned build profile under ADR-0009, secrets-engine fail_closed. Both
conformant, neither a defect, and they disagree about the one case nobody
planned for. Reported as an observation for gate-house's register, not as
a request that either repository change: a PDP does not set a consumer's
stance, and §9.3's two-owner split is our own finding.
The rows are not comparable. ops-warden scopes by security-zone,
secrets-engine by catalog-stage. §6.4 permits both, but the register
cannot then answer what the estate's stance is for a z2 workload. Worth
recording before a third row arrives.
secrets-engine's map defines fail_closed in terms of a durable
ActionAuthorization record, which GH-DEC-2026-005 shelved. The stance is
unaffected -- only the artifact name is stale -- but the file is read at
runtime and pinned by test, so the stale name outlives a comment.
SEPARATELY, T04 IS BLOCKED AND THE REASON IS STRUCTURAL. The task assumed
the ops-warden/tenant-engine/user-engine pattern, where the pin's
default-deny NetworkPolicy admits one approved consumer workload.
secrets-engine has no Kubernetes deployment at all -- it is a CLI. There
is no pod selector to write, and inventing one would repeat the error
corrected in T02. Three possible shapes recorded in the workplan and
raised with them; callerAuth, not the NetworkPolicy, becomes the real
boundary if an operator CLI is the caller, so the FLEX-WP-0016 precedent
does not transfer unexamined.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
Updated by fix-consistency on 2026-09-06:
- update .custodian-brief.md for flex-auth
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
Two real DecisionEnvelopes in examples/secrets-engine/replay/ for
secrets-engine to verify its digest join unchanged: a plain allow
(rotate, empty context) and the dual-control allow (destroy with a valid
approval-claim).
Both are included deliberately. input_claim_digests.context appears only
when the request carries a non-empty context, so a consumer asserting the
field is always present would pass on destroy and fail on rotate. One
fixture would have hidden that.
request_digest, policy_package_digest, registry_snapshot_digest and the
context claim digest are verified identical across two runs and are the
fields to pin. id, decision_time and the lifetime bounds move with the
clock; the README says so rather than leaving a consumer to discover it
by flake. lifetime.ttl is 15m from the package allow_ttl.
Emitted from flex-auth/local in standalone mode, not from a cluster pin.
T03 is progress, not done -- it closes when secrets-engine confirms their
validator accepts the records unchanged.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
approval-engine flagged the class one message earlier -- a contract whose
examples contradict its prose gets implemented as its examples -- and
yesterday's package was a fresh instance of it, committed while flagging
it.
The published rule required context.approval.status == "approved" and
counted context.approval.approvals[].subject_id. Neither field exists.
approval-engine's approval_claim.schema.json has `state` (whose operative
value is `valid`, not `approved`) and carries no approver list at all. The
rule was unsatisfiable: every live destroy would have denied
dual_control_required no matter how good the approval was. It failed
closed, so it was never a hole, but it was policy written against a shape
of our own devising rather than a published one.
The rule now consumes valid_now from the real claim, guarded on kind and
issuer. valid_now is the summary predicate that already folds in the
distinct-approver threshold, with reason_code insufficient_approvers for
a claim that failed it -- so this is also the correct layering, not just
the correct shape. Counting approvers here is exactly the duplication
GH-DEC-2026-005 removes; the compensating property is reconstructability
at the issuer under 9.6, which is approval-engine's.
Recorded as a correction section in the package and the vocabulary doc
rather than quietly rewritten. 25 Rego tests and 29 fixtures pass,
covering insufficient_approvers, consumed, revoked, approved-but-not-yet-
valid, foreign issuer, and wrong kind.
Two things the package deliberately does not do, both now written down:
it does not compare binding.pdp_digest, because the request digest is
computed after policy evaluation and a Rego rule cannot see it; and it
makes no cross-check that the claim was approved for this action and
target, because the claim's binding uses approval-engine's vocabulary and
no mapping between the two is published. Inventing one would silently
accept a claim approved for something else. Both belong to the PEP until
a mapping exists, and that is worth closing before SECRETS-WP-0007-T04
makes destroy reachable.
Also swept the other published fixtures on approval-engine's reasoning.
One more instance: the inner decision in examples/caring/action_authorization.json
declared contract_version flex-auth.decision-record.v1 while its
provenance omitted policy_package_digest, registry_snapshot_digest, and
input_claim_digests -- all published contract fields since 2026-09-02.
Completed. The remaining example context vocabularies are consumer-owned
and match their integrations.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
Updated by fix-consistency on 2026-09-06:
- update .custodian-brief.md for flex-auth
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
secrets-engine delivered the action vocabulary T01 asked for: twelve
actions read out of cli.py, not the example vocabulary. The gate earned
its keep -- four would have been inferred wrongly, and revoke, the
obvious thirteenth, does not exist as an action at all. It gates as
deactivate, which is also reached from lifecycle deactivate.
docs/secrets-engine-action-vocabulary.md records the list and the four
traps. examples/secrets-engine/ carries the package, both manifests, a
loadable registry snapshot, 26 fixtures, five check requests, and a
README. validate -kind policy is valid with 22/22 Rego tests and 26/26
fixtures.
allow_ttl is 15m, stated in the package rather than inherited from the
engine default: these decisions authorize live secret operations, so the
reliance window belongs where a reviewer sees it. Per FLEX-DEC-2026-004
it is authority to issue, not authority to keep using what the operation
produced.
destroy is the dual-control case, gated on a context.approval claim
marked approved with two distinct approvers, repeated entries counting
once. flex-auth checks what the claim says and deliberately does not
re-derive its temporal validity, signature, or supersession -- those are
approval-engine's to assert and the PEP's to verify against the live
claim, per the split accepted in FLEX-DEC-2026-006. It stays in the
package though its handler raises before the gate, so the rule is
reviewed and fixtured before SECRETS-WP-0007-T04 opens the path.
Following the FLEX-WP-0010-T02 precedent the denial ladder has no
action_not_granted branch: one subject holding all twelve actions could
never reach it, and a rule that cannot fail reads as control that is not
there. Registering a second calling identity is the revisit trigger.
Not deployed. No flex-auth-secrets-engine pin exists yet (T04), so their
policy pin stays unset and fail-closed until T05.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
approval-engine filed APPROVAL-IN-0002: secrets-engine built its PEP
validator against our ActionAuthorization schema, pointed it at
GET /v1/approvals/{id}/claim, and it rejects every response. Both
envelopes declare schema_version 0.1, so it fails late and reads like an
approval-engine outage rather than a contract mismatch.
FLEX-DEC-2026-006 accepts the deferral and argues against flex-auth's own
proposal. The composed object had the PIP republish our decision, which
crosses the same layer boundary we invoked to decline authentication
evidence and to win section 17's schema. The claim-plus-DecisionEnvelope
split drops no check; each verification lands on the layer that owns it.
approval-engine asked, before the decision, whether the open G3 finding
argues for ratifying now. It does not: G3 is already closed the other
way. FLEX-WP-0019 added lifetime to the DecisionEnvelope itself, required
on every allow by schema conditional, published 2026-09-02. The trigger
resolved by adding a field rather than by composition, so the decision
stands alone and needs no bundle.
The provenance.authority == state-hub constant is our defect and is
fixed at source. It came from examples/caring/action_authorization.json,
which contradicted the same contract's ownership section. That fixture
now names approval-engine as the approval fact's authority and flex-auth
as the decision's, and its stale secrets-engine.lifecycle pin is
corrected to the reserved coordinate from FLEX-DEC-2026-005.
The contract doc and schema are marked deferred-not-withdrawn so no
other consumer builds a validator against them. The execute-time half is
untouched: /v1/check, binding, the canonical digest, and
flex-auth.decision-record.v1 stay published.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
Two cross-repo questions arrived in the flex-auth inbox and both are
answered as decision records rather than as prose in a message.
FLEX-DEC-2026-004 answers ops-warden WARDEN-WP-0034-T05. A decision
lifetime shorter than the SSH certificate TTL is meaningful, but only as
authority to issue, never as authority to use an already-issued
certificate. The pre-sign gate is the only consumer of the shorter
lifetime: no replay past expires_at, fresh Check per sign. The lever that
shortens effective access is the requested TTL as a policy input, which
is already deployed as the ttl_out_of_bounds deny. ops-warden's section
9.7.2 window through certificate TTL is correct as written and correctly
owned by the PEP; flex-auth does not want that residue moved to the PDP.
docs/decision-input-freshness.md gains the same boundary as published
contract text, so the ruling is not only in the decision log.
FLEX-DEC-2026-005 answers secrets-engine. A real policy package is
expected and flex-auth authors it here as it does for every consumer; the
reserved coordinate is secrets-engine.catalog-lane.lifecycle v1 and it
does not exist yet. Their choice not to default the pin was correct and
is endorsed explicitly. POST /v1/check is deployed but has no
estate-wide address by design -- per-consumer cluster-local pins with
default-deny ingress -- so their 2026-09-06 probe found the design
working, not an outage.
FLEX-WP-0021 carries that work: obtain the real action vocabulary from
secrets-engine, publish the package with fixtures, confirm the digest
join against a real decision record, then stand up a
flex-auth-secrets-engine pin in warn without moving the other two pins.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JTbVXpEiXA7mNJVpDnEPcB
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 412054@bnt-lap001
Assistant-Session: 3968fae1-8d59-4209-9bd6-c22594b8ab19
Close the remaining PDP obligations: mechanical layer declaration check,
registry-snapshot digest in provenance, explicit allow TTL, per-input-class
freshness deadlines, and the published decision-record contract. Document
the canonical request digest as the §6.4.2 replay test.
Assistant: grok
Assistant-Session: 01a06256-fb71-7102-b3a9-27e6734257d0
Updated by fix-consistency on 2026-09-03:
- update .custodian-brief.md for flex-auth
Assistant: grok
Assistant-Session: 01a06256-fb71-7102-b3a9-27e6734257d0
C-21 flagged the ready workplan as possibly stale: reviewed against 5753b47,
HEAD had moved. Checked rather than assumed — nothing since 5753b47 touches Go
or schemas. Every commit between is documentation or work-record only, and the
last implementation change is d402408 (2026-08-23).
reviewed_against_commit now names the implementation state reviewed rather than
the HEAD at review time, so it moves when the code moves, which is when the
review needs redoing. C-21 compares against HEAD and cannot distinguish a doc
commit from a code one, so it will keep warning softly on this plan; the record
is accurate and that is the trade taken deliberately.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012sgN4GH5ZYT8pJVkCR6dcP
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 4014348@bnt-lap001
Assistant-Session: a993abda-65a0-4ea8-8ccd-0fcd78c92ac0
C-34 flagged status=ready without a DoR assessment. The warning was right: the
plan claimed repo-state review with no review metadata, and work-item-dor.md
prefers DoR-Failed for a ready claim in that state.
Adds reviewed_at, reviewed_against_commit (5753b47 — the code state the
alignment review actually checked each obligation against), the origin pointer
to the review, and quality_dor: DoR-Ok with the criteria walked in the note.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012sgN4GH5ZYT8pJVkCR6dcP
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 4014348@bnt-lap001
Assistant-Session: a993abda-65a0-4ea8-8ccd-0fcd78c92ac0
Updated by fix-consistency on 2026-08-29:
- update .custodian-brief.md for flex-auth
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 4014348@bnt-lap001
Assistant-Session: a993abda-65a0-4ea8-8ccd-0fcd78c92ac0
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012sgN4GH5ZYT8pJVkCR6dcP
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 4014348@bnt-lap001
Assistant-Session: a993abda-65a0-4ea8-8ccd-0fcd78c92ac0
The standard was accepted at v0.7 on 2026-08-29. Four of flex-auth's review
findings are in the accepted text: 9.3's two-owner split, 6.4.2 scoped to the
decision's own binding with our canonical request digest as its mechanical test,
9.7.2 split by role, and 17 moving the decision-record schema to access-engine.
INTENT.md
- Machine-readable layer declaration in frontmatter (layer: Engine, role: PDP),
which section 11 requires and we did not have. audit-core noted our
declaration was legible only by following the decision trail.
- PDP failure semantics stated: our outage is consumer residue, not input
degradation; fail-open is not expressible by a PDP at all.
- Four owned obligations added: the decision-record schema as our contract, the
request digest as the published replay test, a lifetime on every allow, and
visibility deadlines per input class.
- A Layer Conformance section stating the state honestly: conforming with one
declared gap, no Tooling client, not PEP-shaped.
- Vocabulary correction: earlier text dropped "control plane" as Staff
vocabulary. Section 8 binds it to the Engine layer, which is why kings-guard
was asked to release it. The term is ours; we prefer "decision engine" for
precision, not boundary.
SCOPE.md
- Layer and role in the one-liner; the four obligations In Scope; five
boundaries established in review but never written down Out of Scope.
- Three capability blocks marked planned for workplans completed in May are now
current; two blocks added.
- Superseded ADR-0006 citation corrected to ADR-0009, which retires the global
flag outright rather than deferring it.
history/2026-08-29-layer-model-v0.7-alignment-review.md checks each obligation
against the code and finds six gaps. FLEX-WP-0019 closes them, with T02 before
T04 because a visibility deadline for registry-borne facts is unfalsifiable
until provenance can identify the snapshot a decision read.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012sgN4GH5ZYT8pJVkCR6dcP
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 4014348@bnt-lap001
Assistant-Session: a993abda-65a0-4ea8-8ccd-0fcd78c92ac0
