Updated by fix-consistency on 2026-09-21:
- update .custodian-brief.md for secrets-engine
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 63291@bnt-lap001
Assistant-Session: 8bd77868-ca68-4f49-bb1e-d539ecc0d703
railiance-platform e82bb289 recorded against SECRETS-WP-0006-T05 (explicit wait
on T04 serving; CCR-2026-0003 is provenance only). 29cccf8a recorded against
SECRETS-WP-0008-T06, including the open tenant:coulomb vs tenant:platform
question for the service JWT, left for an owner session. railiance-clock's
review request opened as SECRETS-IN-0003. The intelligence-radar messages are
superseded by SECRETS-WP-0010-T03 (done 2026-09-16) and answered by pointer.
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
Verified against gate-house's committed ruling (decisions/decisions.md,
GH-DEC-2026-017) and amendments A9-A13, then ops-warden's reference change set
(a70f559, wiki/playbooks/netkingdom-layer-declaration.md). They agree.
layer.yaml: standard_version removed; derived: true and derived_from:
INTENT.md added; declared_by kept. INTENT.md frontmatter never carried
standard_version, but its standard: value was a version-pinned path; it is
de-versioned as the reference instance did. No layer value is re-spelled:
INTENT.md still says Engine and layer.yaml still says engine.
The checker changes in the same commit because it listed standard_version as
a required key: removing the field alone would have made a conforming
declaration exit 2 MALFORMED. It now reads INTENT.md as the governing form,
requires the derived marking, rejects a returning standard_version in either
form, checks both layer values against the closed four-token vocabulary
(Taxonomy included) after an ASCII fold, and reports a post-fold disagreement
between the forms as a finding rather than resolving it by precedence.
Tests assert the fold, not per-file spelling, and cover fold agreement, a
real disagreement, the closed vocabulary and a returning version. Full suite
430 passed.
role:, pep-stance.yaml and schema_version are untouched (not ruled). Still
open: where the removed version lives in a derived conformance record; asked
of gate-house by ops-warden (4220413a), followed rather than chosen here.
Closes the SECRETS-WP-0008 note that waited on the reference form.
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
Updated by fix-consistency on 2026-09-21:
- update .custodian-brief.md for secrets-engine
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 63291@bnt-lap001
Assistant-Session: 8bd77868-ca68-4f49-bb1e-d539ecc0d703
Consumer-surface survey found exactly one live repository path, a local
checkout path in docs/approval-service-auth.md. Every other flex-auth string
is a runtime or contract name FLEX-DEC-2026-013 keeps. Held open until the
rename lands rather than documenting a path that does not exist yet.
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
ops-warden and railiance-platform asked, independently, which OpenBao location
`secrets-engine exec --catalog whynot-design-npm-publish` reads. Answered from
this repository's own code and catalog with no OpenBao read and no value.
It reads `secret/coulomb/whynot-design/npm/publish` — the legacy, ungoverned
duplicate. `_fetch_value` concatenates the catalog's `mount` and `path` with no
override or fallback, so the proven pilot published from the duplicate and the
lowercase `npm_token` field is the field there. ops-warden's front door names
this repository as exec_owner, so it currently routes callers at a path no CCR
covers.
Recorded in SECRETS-WP-0006 with what a move to the governed lane requires, and
flagged that the duplicate must not be destroyed until the lane moves. Whether
the two locations hold the same value is a value comparison and is not answered
here.
Also records GH-DEC-2026-017 against SECRETS-WP-0008: INTENT.md governs, the
sidecar is derived, the vocabulary is case-insensitive so nothing is re-spelled,
and standard_version comes out of layer.yaml once ops-warden updates the
reference form this repository copied.
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
Set flavor on open workplans from origin/prose/status. Copy existing
depends_on aliases only. Do not promote residuals.
Assistant: grok
Assistant-Session: 01a09dc1-b21e-77e1-919e-fcad2f82b267
SCOPE.md is the capability boundary other agents read, and three of its
claims had gone stale — one of them describing checks that were deliberately
removed months of decisions ago.
- It said the engine "builds and validates the flex-auth ActionAuthorization
profile", including State Hub authority and an independently required
distinct-approver threshold. All three are wrong: ActionAuthorization is
deferred and never ratified (FLEX-DEC-2026-006) and nothing validates it,
the State Hub authority constant was removed because State Hub is a read
model with no runtime approval authority, and the approver threshold is
folded into valid_now by the issuer rather than re-checked here. Replaced
with the actual two-artifact split from GH-DEC-2026-005, including the
reduction in what this engine verifies alone, stated rather than buried.
- It said the access-engine serving endpoint does not exist. It does, and
step 2 is proven against it. Step 1 is the unserved half.
- The layer-model table row repeated the ActionAuthorization framing.
Also records the structured-correspondence rule and the tenant requirement
in the capability list, and adds responder authentication to Not Implemented
— it is a real gap in what this engine can promise, not merely flex-auth's
open work.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E4tNMAYcSQmZWUE4wqP4ij
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715726@bnt-lap001
Assistant-Session: 80a42b32-cba6-4b23-8be0-68819b1a6092
SECRETS-WP-0008-T02 still said access-engine Check was unreachable. That is
no longer true and the note read as current, so it is corrected rather than
left to mislead: Check is reachable through the owner-documented access
path and step 2 is proved against decision:0f9c98f14545c42d, a real v2
allow. Two defects that only a real request could expose -- the missing
tenant and the unsatisfiable digest join -- are fixed, so the task's own
acceptance line about failing closed on wrong digest and expired lifetime is
now exercised against a genuine envelope rather than a fixture.
What remains for that task is step 1 alone: approval-engine must serve the
claim endpoint, and APPROVAL-WP-0002-T03 is still wait with no deployed base
URL. One external dependency, not the two previously named.
SECRETS-WP-0006-T06: asked railiance-platform which KV location backs the
whynot-design npm publish lane (hub message 546403e4). The question is
narrowed to the path, since the endpoint agrees and ops-warden's
NPM_AUTH_TOKEN claim resolved here as a category error -- that is the
resolved injection env var, not a KV field name. Catalog left unchanged:
rewriting a proven production lane pointer from an inbox claim is the
unverified custody mutation this task exists to prevent.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E4tNMAYcSQmZWUE4wqP4ij
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715726@bnt-lap001
Assistant-Session: 80a42b32-cba6-4b23-8be0-68819b1a6092
Updated by fix-consistency on 2026-09-07:
- update .custodian-brief.md for secrets-engine
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715726@bnt-lap001
Assistant-Session: 80a42b32-cba6-4b23-8be0-68819b1a6092
The live proof in 03c0569 showed validate_decision_envelope rejecting every
real allow. The evaluator normalizes before hashing -- the request tenant is
copied onto subject and resource, and a registry hit copies type, tenant and
selected attributes onto the refs -- so binding.request_digest covers
material we never sent. Byte-equality against our unenriched request was
unsatisfiable, not merely mismatched.
THE RULE WAS ALREADY PUBLISHED. flex-auth's canonical-request-digest.md
section "Normalization" states the enrichment and tells consumers what to do
instead: compare structured binding fields to the proposed action, treat
request_digest as the evaluator's statement of what it hashed, and recompute
independently over the tuple the binding carries. I raised this with them as
an unpublished gap and asked them to pick between three shapes; it was in
their contract already and the answer was the first of the three. Nothing
was blocked on them, and this follows the published rule rather than one I
inferred.
- _require_binding_corresponds: everything we proposed must survive
unchanged -- tenant, action, context, subject.id/type,
resource.id/type/system, and every attribute we sent.
- Enrichment may add only type, tenant, attributes. Any other added field is
refused, and an enriched tenant must be the request tenant, so a
cross-tenant binding cannot arrive wearing our request's clothes.
- request_digest is still verified, now against binding_tuple(binding) for
self-consistency rather than against material we never sent.
- The envelope's top-level subject/resource get the same rule; they are
enriched too.
Proved against the artifact: the real decision:0f9c98f14545c42d now
validates, and the unrefreshed envelope is refused on lifetime -- reaching
the lifetime check at all is the evidence the binding checks pass on a real
decision. Negatives cover a restated resource.attributes.stage, a foreign
subject.tenant, and an unexpected enrichment field.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E4tNMAYcSQmZWUE4wqP4ij
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715726@bnt-lap001
Assistant-Session: 80a42b32-cba6-4b23-8be0-68819b1a6092
Adoption asked for by flex-auth (FLEX-WP-0021-T05) and glas-harness
(GLAS-WP-0015), plus the first real decision this engine has obtained from
the deployed pin -- which found a defect the fixtures could not.
ACCESS PATH. require_supported_pdp_address refuses in-cluster Service names
and any non-loopback host. This is no longer a unilateral call: the owner
path is documented as loopback kubectl port-forward over the authenticated
Kubernetes API, which is what authenticates the responder transitively
(FLEX-DEC-2026-010). A Service name from a workstation does not fail, it
resolves through the DNS search suffix to an unrelated public host, and
since decision records carry no signature, a responder knowing the
published package and version can return an allow that passes every check
we make. Fail-closed protects against a PDP that is absent, not one that
lies. The guard runs before the token is read, so a misdirected request
cannot leak it; a test pins that ordering.
LIVE PROOF. Minted a 10-minute TokenRequest token (audience flex-auth, SA
secrets-engine/secrets-engine, mode 0600 outside the worktree, shredded
after), forwarded to the named pod, and sent a real CheckRequest for
glas-claude-agent-dev-anthropic. Result: allow, catalog_lane_policy_matched,
served by v2 (sha256:bd11c5fe...) -- so the redeploy flex-auth flagged as
outstanding has landed and the pin no longer serves the tenant-blind v1.
Our tenant fix is confirmed against the real service: binding.tenant is
tenant:platform.
THE DEFECT IT FOUND. The evaluator enriches from its registry before
hashing -- subject gains attributes and tenant, resource gains tenant --
so binding.request_digest is over material we never sent and cannot
reproduce. validate_decision_envelope rejects every real allow.
Every replay test passes because _request_from() rebuilds the request out
of the binding, i.e. the enriched form: a self-consistent fake agreeing
with itself, which hid this through three rounds of digest work. Third
time a real artifact has beaten a fake in this integration.
NOT FIXED, DELIBERATELY. Rejecting a valid allow is wrong in the safe
direction. Which fields may be enriched is flex-auth's contract to publish;
inferring it means accepting a binding that differs from our proposal in a
way we decided was benign -- the fail-open shape GH-DEC-2026-008 rejected
for vocabularies and FLEX-DEC-2026-007 for digests. Raised with them.
Tenant question closed by operator decision 5ed3fb35: tenant:platform
exactly, and service_auth.TENANT stays tenant:coulomb because the two
identity layers are to remain distinct. Declining to author that mapping
was right -- the answer was neither reading offered.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E4tNMAYcSQmZWUE4wqP4ij
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715726@bnt-lap001
Assistant-Session: 80a42b32-cba6-4b23-8be0-68819b1a6092
Updated by fix-consistency on 2026-09-06:
- update .custodian-brief.md for secrets-engine
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715726@bnt-lap001
Assistant-Session: 80a42b32-cba6-4b23-8be0-68819b1a6092
Probing the handed-over Service DNS name from the workstation found that
every *.svc.cluster.local name resolves here to one unrelated public
address via the ad.binect.de search suffix -- including names of services
that do not exist, which proves it is suffix expansion and not a record.
Pointing SECRETS_ENGINE_PDP_URL at the Service name would send the
CheckRequest body (subject, tenant, lane ids, stage, field names, purpose)
and the static Bearer token to that host. Decision envelopes are
structurally validated but not signed, so a responder knowing the package
and version can return a well-formed allow; the fail-closed posture
assumes the PDP is the PDP.
Not mitigated here: choosing the transport control is the owners' call,
not this consumer's. Recommended to FLEX-WP-0021-T05 that the handover be
a trailing-dot FQDN or explicit address, with the response channel's
authentication stated. Our pin stays unset, so the hazard is theoretical.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E4tNMAYcSQmZWUE4wqP4ij
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715726@bnt-lap001
Assistant-Session: 80a42b32-cba6-4b23-8be0-68819b1a6092
build_action_request emitted no tenant field at all. The deployed
secrets-engine.catalog-lane.lifecycle v2 package reads
known_tenant := "tenant:platform"
request_tenant := object.get(input, "tenant", "")
so an absent tenant is not an ignored field, it matches the wrong_tenant
first-denial branch. Every gated action this engine sent would have been
denied -- and the omission also produced a request_digest that could match
no correctly issued decision, since tenant is hashed material. That is the
same class of defect as hashing excluded fields, arriving from the other
direction, and again only a real artifact exposed it.
Found by answering the GLAS-WP-0015 tenant-alignment question instead of
assuming the values lined up.
- REQUEST_TENANT is pinned against the vendored allow envelopes, so a
package retenanting fails a test rather than denying production.
- An empty tenant is refused at build time.
- Accepted policy version moves v1 -> v2. v1 had no tenant rule and failed
open: a rotate under tenant:coulomb returned allow against the deployed
package. flex-auth superseded rather than amended it, because a fail-open
correction has to be visible as a version change. A test pins that a v1
decision is refused.
- Vendored decision_wrong_tenant_deny.json as the denial evidence glas
asked for, with tests that we refuse it on effect before anything else
and that a deny legally carries no lifetime.
docs/tenant-alignment.md states the three tenant values as this repo holds
them. It does not resolve the JWT/store mapping: service_auth.TENANT is
tenant:coulomb, which is exactly the value the package denies. That is
either two layers sharing a namespace format or one wrong constant, and
picking between them without an owner ruling is the fail-open shape
GH-DEC-2026-008 rejected for action vocabularies. Both constants stay as
they are, deliberately not unified.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E4tNMAYcSQmZWUE4wqP4ij
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 715726@bnt-lap001
Assistant-Session: 80a42b32-cba6-4b23-8be0-68819b1a6092
