Reachability per endpoint, setup without the window-scoped write role (no Kubernetes or directory
client), then every preflight check (admission, executor, source, trust, registry, window,
approval) with its phase; no consume, emit or write. Exit 0 ready, 4 not. preflight() is now the
same named checks, stopping at the first. Setup reasons keep the clients' fixed texts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 943911@bnt-lap001
Assistant-Session: fc322940-1c8e-4bed-ab39-86cec99f23a0
executor-lanes.py: the identity phase distinguishes openbao_unreachable, openbao_sealed and
platform_admin_login_required, with a one-line hint; LaneError keeps its fixed code. executor.py:
approval_entry:<approver_not_listed|not_human|not_aal2|self_approval>, never the subject.
Note: executor.py changed, so the next admission pins a new image.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 943911@bnt-lap001
Assistant-Session: fc322940-1c8e-4bed-ab39-86cec99f23a0
Two live runs of BEA-2026-0002 ended setup:URLError while four read-only diagnostic pods (same
image, labels, mounts and config) reached every endpoint. Likely cause: a fresh Job pod reaches
OpenBao before the sso/approval-engine ingress rules admit it (NetworkPolicy is programmed per
destination). Now:
- build waits until OpenBao, KeyCape, approval-engine, audit-core and LLDAP all answer HTTP
(one shared 30 s budget); an unreachable one is named: setup:<name> not reachable.
- a preflight error is a refusal named preflight_error:<Type> (preflight has no effect).
- an error in the consume read-back is outcome unknown, consume_unconfirmed:<Type>, alerted;
before, any exception after setup was reported as 'refused, no effect' even after consume.
- any other escaping error after setup is outcome unknown (exit 5), never 'no effect'.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 523928@bnt-lap001
Assistant-Session: 6c737afb-9c14-4152-9f02-477214b9c350
- cli: SIGTERM/SIGINT (kubectl delete job) sets the stop flag; the runner stops at the next
step boundary and reverts. A stop before consume refuses and spends nothing.
- posture feedback v0.1 section 4a: --receipts reads bounded-executor events through the
receipt validator: execution-receipt-missing, -not-completed, -reverted, -budget-pressure,
and the aggregate execution-recurring-revert. Owners come from the authoritative
tools/posture-feedback/execution-owners.yaml, never inferred.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 523928@bnt-lap001
Assistant-Session: 6c737afb-9c14-4152-9f02-477214b9c350
- canon/schemas/execution-attribution-receipt_v0.1.schema.json: the bounded-executor profile
of the PCC v0.2 receipt, both audit-core events (started, receipt), every object closed.
- tools/execution-attribution-receipt: EAR-1..EAR-10 validator, fixtures generated from the
real executor (completed, reverted, completed-on-retry) with a drift check.
- executor: started carries receipt_version, digests, steps_planned and data.window.not_after
(audit-core AUDIT-WP-0018-T03 request); receipt carries receipt_version and initiator.iss
(configured approver_issuer, else unknown).
- bounded-executor/exercise_audit_sender.py: real executor and client against audit-core's
real ingestion and the Git scope overlay: 202, duplicate 200, conflict 409 as a defect,
wrong source/tenant/token refused and not stored, a receipt-less start flagged by its own
window.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 523928@bnt-lap001
Assistant-Session: 6c737afb-9c14-4152-9f02-477214b9c350
The exercise runs the real request phase, KeyCape client credentials, preflight and consume
against KeyCape's pinned image and approval-engine's source with synthetic keys.
It found that approval-engine stores an entry's assurance as the verified token's JSON object
(KeyCape assuranceClaim), so the executor's comparison with a bare "aal2" refused every real
founder approval. entry_is_aal2 now requires level aal2/aal3 with mfa true and fails closed
otherwise. Fakes use the real shape.
The request-phase tests freeze the clock inside the example admission's window (they had
expired on 2026-11-03T11:00Z).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 523928@bnt-lap001
Assistant-Session: 6c737afb-9c14-4152-9f02-477214b9c350
CLI entry point and approval-engine/audit-core clients; events carry
occurred_at (audit-core envelope); hash-locked image with pinned lldap
stage builds locally; image workflow and executor tests in CI; pilot
example pins the executed files. rapp reshaping still open.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 994874@bnt-lap001
Assistant-Session: 6017c416-d619-47eb-bd1a-f670a02d7d87
operations.py: OpenBao KV v2 CAS, LLDAP self-rotation (forward and revert),
ExternalSecret force-sync and status, rollout restart (RolloutFailed is a
definite failure), pilot checks. synthetic_pilot.py on the T08 stack plus
OpenBao 2.5.4 with CCR-2026-0058's policy: rotation completed; a one-off wrong
delivery reverted in declared order with the earlier value restored.
Runner: revert steps run in declared order (BEA-10), definite failures revert,
unknown outcomes stop, receipts carry error types. Found and corrected (in
railiance-platform) CCR-2026-0058's required_parameters refusing reads.
Packaging moved to new T13. 29 executor tests.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 994874@bnt-lap001
Assistant-Session: 6017c416-d619-47eb-bd1a-f670a02d7d87
Consumes one admission once through the approval-engine carrier after an
effect-free preflight (validity, source pins, trust, window vs budget,
binding digest, actor, human aal2 approvers), emits run.started, runs steps
with settle-timed checks, reverts on failure with per-revert checks, retries
within attempts, stops on unknown outcomes, honours budget and kill switch,
and emits one value-free receipt (PDP not-evaluated). 22 tests with fakes.
BEA-9 extended to revert steps; examples regenerated.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 994874@bnt-lap001
Assistant-Session: 6017c416-d619-47eb-bd1a-f670a02d7d87
