CURRENT VALIDATIONS

Validated in isolated v1.1 qualification environments. Evidence applies only to the stated scope and environment.

Traffic delivery, restart and replay recovery

VALIDATED · 2026-09-15

24/24 transport attempts -> 24 target effects, 0 lost; restart/recreate/replay recovery; security negatives rejected

Scope: isolated v1.1 qualification

Evidence ID: traffic-delivery-recovery-v1

Control-plane-independent established delivery

VALIDATED · 2026-09-16

24 accepted events fully accounted during 30-minute complete control-plane blackout; 22 delivered, 2 durable-terminal during nested target outage per bounded-retry contract; 0 manual repairs; DP/EP restart recovery PASS

Scope: isolated v1.1 qualification, 30m blackout

Evidence ID: traffic-cp-independent-v1

Fabric publication and convergence across outage and restart

VALIDATED · 2026-09-15

pipeline activation published through real Control->Fabric over strict mTLS to durable DB; plane consumed; generation convergence; fabric/control restart recovery; stale/tampered rejected

Scope: isolated v1.1 component-live qualification

Evidence ID: fabric-publication-convergence-v1

Trust live boundary enforcement (negative matrix)

VALIDATED · 2026-09-15

certificate, principal, environment, tenant, generation and digest negative cases rejected over live network interfaces

Scope: isolated v1.1 component-live qualification

Evidence ID: trust-live-boundary-negative-matrix-v1

Evidence is published only after validation and disclosure review. Superseded records remain in history but are not used as current proof.

Evidence

AI-readable evidence

Curated, machine-readable artifacts for reviewers and LLM crawlers. Proof status is scoped; local, mock, and staging results do not equal production readiness or customer-ready operation.

Global evidence baseline scope: sandbox qualification topologies (frozen-green baselines and verified-current capabilities listed below). Local, mock, and staging results do not equal production readiness.

24-hour qualification runs

Survival qualification of a release candidate over a continuous 24-hour window in an isolated local environment. PASS/FAIL semantics apply to the 24-hour window only; the post-24h period is a lifecycle observation tail and never changes the result. Rendered mechanically from the sanitized projection of the private evidence bundle — never hand-copied.

CP24-2026-09-rehearsal — 24-hour survival qualification (rehearsal)

QUALIFICATION_24H_SURVIVAL_PASS · isolated local qualification environment · started 2026-09

What was tested: continuous 24-hour unattended operation of the delivery path with synthetic journeys, identity lifecycle active, and fail-closed security negatives exercised.

What passed: 666 of 666 synthetic deliveries verified at the target (664 immediate, 2 write-time pending later verified); ZERO_UNEXPLAINED_DELIVERY_LOSS; ZERO_SEMANTIC_DUPLICATE_EFFECTS; ZERO_MANUAL_REPAIR; SECURITY_NEGATIVES_HELD (fail-closed negatives exercised during the window).

Post-24h tail: POST24_BOUNDARY_NOT_OBSERVED (observation only — does not affect the qualification result).

What was NOT proven: no production proof; no automatic identity rotation proof; no post-expiry reconnect proof.

  • Limitation: active probe stream ended near the 24-hour boundary; the exact post-boundary flow-stop time was not observed
  • Limitation: one long-lived relay session persisted for the whole window, so transparent identity replacement and post-expiry reconnect were NOT proven by this run
  • Limitation: the nearest identity boundary occurred after evidence harvest; no certificate rotation was observed
  • Limitation: environment: isolated local qualification lab, not cloud or production infrastructure

Methodology: zen-mgmt/cp24-runbook/1.0.0 · public evidence digest: sha256:95aa2f397abc70a6…

Product proof — packet-fed

Per-product proof rendered mechanically from digest-verified fleet-bus packets (9 packets, sha256-bound below). Claims are never hand-copied: a generator derives them from the same facts the rule engine sees, and a consistency check re-derives every load-bearing field from the packets — a hand edit fails the build's validation battery. Verifier classes are carried verbatim as the producing lane labeled them (SELF_REPORTED ≠ INDEPENDENT_ACCEPTED); this page never upgrades a label.

Zen Fabric (pre-SVID plane security)

source packet: daedalus.bus-report · generation: C126/C127 fact vocabulary; packet generated bus-live

  • pre_svid.route_classification LIVE PROVEN (PRODUCING LANE)

    route census guard green (4db79e6)

    evidence: daedalus.bus-report#pre_svid · verifier: producing-lane labeled (Daedalus); independent re-verification welcome

  • pre_svid.mode_b_key_binding LIVE PROVEN (PRODUCING LANE)

    admission binding inside grant signature (f09c293)

    evidence: daedalus.bus-report#pre_svid · verifier: producing-lane labeled (Daedalus); independent re-verification welcome

  • pre_svid.private_key_custody STRUCTURAL PROVEN

    no private-material fields

    evidence: daedalus.bus-report#pre_svid · verifier: producing-lane labeled (Daedalus); independent re-verification welcome

  • pre_svid.one_time_consumption LIVE PROVEN (PRODUCING LANE)

    replay 409 across 3 windows

    evidence: daedalus.bus-report#pre_svid · verifier: producing-lane labeled (Daedalus); independent re-verification welcome

  • pre_svid.replay LIVE PROVEN (PRODUCING LANE)

    evidence: daedalus.bus-report#pre_svid · verifier: producing-lane labeled (Daedalus); independent re-verification welcome

  • pre_svid.revision_binding LIVE PROVEN (PRODUCING LANE)

    stale-admission refusal + revision currency at mint

    evidence: daedalus.bus-report#pre_svid · verifier: producing-lane labeled (Daedalus); independent re-verification welcome

  • pre_svid.protocol_versioning GOLDEN VECTORS

    19 machine-executable (f6fadd2)

    evidence: daedalus.bus-report#pre_svid · verifier: producing-lane labeled (Daedalus); independent re-verification welcome

  • pre_svid.downgrade_refusal GOLDEN VECTORS

    v0/future typed-refused

    evidence: daedalus.bus-report#pre_svid · verifier: producing-lane labeled (Daedalus); independent re-verification welcome

  • pre_svid.compatibility GOLDEN VECTORS

    N coexistence proven

    evidence: daedalus.bus-report#pre_svid · verifier: producing-lane labeled (Daedalus); independent re-verification welcome

  • pre_svid.resource_bounds CODE PROVEN

    MaxBytesReader + bounded admission + grant TTL

    evidence: daedalus.bus-report#pre_svid · verifier: producing-lane labeled (Daedalus); independent re-verification welcome

Pre-SVID facts describe the admission/grant plane before workload identity issues; they are not an SVID-issuance or production-live claim.

Zen Remediator (typed refusal surface)

source packet: scout.r041-negatives · generation: R041 typed negatives battery (10 negatives, all expected-typed)

  • typed_negatives_battery PASS (TYPED REFUSALS)

    All 10 probes typed-refused with no silent acceptance (9 exact expected status + the N10 three-phase typed sequence: expiry refusal before the approval check, then the live non-resurrection 409).

    evidence: scout.r041-negatives (statuses mechanically compared to expectations at generation) · verifier: producing-lane labeled (Scout R041); statuses re-verified by the consistency check

  • fail_closed_transport LIVE PROVEN (PRODUCING LANE)

    Unconfigured transport fails closed with a typed 503 (DEPENDENCY_UNAVAILABLE) — R041-BROKER-SEAM-OWNERSHIP: the lawful current state, never a silent dead-endpoint wire.

    evidence: scout.r041-broker-seam (typed finding + evidence chain) · verifier: producing-lane labeled (Scout R041); TYPED-BLOCKED remains the lawful posture

Typed refusal behavior is a sandbox/staging-scope observation; it is not a production-hardening claim.

Zen Mon (observation truth)

source packet: mon.bus-report · generation: mon bus report generated undefined

  • reprovision_wake IMPLEMENTED · TEST-PINNED

    Exactly one estate re-provision wake per wiped-to-live boundary; stale post-wipe greens re-typed to UNKNOWN; typed absences preserved verbatim.

    evidence: mon.bus-report#programs.A_reprovision_wake · verifier: producing-lane labeled (Mon), test-pinned

  • presvid_probe_families FIXTURE PROVEN · WAKE-GATED

    8 pre-SVID probe families fixture-proven and wake-gated — the re-provisioned estate is born observed.

    evidence: mon.bus-report#programs.B_presvid_probes · verifier: producing-lane labeled (Mon)

  • self_observability GAPS CLOSED

    zen-mon closes its own DASHBOARDED / RUNBOOKED / RECOVERY_PROVEN gaps; the recovery drill ran against the actual wiped estate (absence-proof receipts).

    evidence: mon.bus-report#programs.C_self_observability · verifier: producing-lane labeled (Mon)

Observation coverage is estate-scoped; absence of observation is always rendered truthfully (no fabricated live data).

Zen Platform console estate (current generation)

source packet: platform.staging-roll-p135 · generation: staging roll at daf30b04

  • serving_generation DIGEST-BOUND DEPLOYMENT

    Console front p135a (sha256:ab278…) 2/2 and BFF p135a-b (sha256:2af9d…) 1/1 built from a clean git export of the pushed HEAD and deployed receipt-bound; live smoke / 200, /console/ 200, auth gate 401.

    evidence: platform.staging-roll-p135 (candidate receipts + live probes) · verifier: SELF_CHECKED (platform lane); independent qualification welcome

Serving generation is a staging fact; READY replicas are not a website-health or production claim.

Delivery estate domains

Estate-level delivery domains as validated by the independent Collector lane (collector.go-nogo-v7); every artifact was hash- or HEAD-checked by Collector. Verdict posture only — decision deadlines and internal gates are not published.

CAPACITY

GO (typed, UI/API plane)

  • P134 overnight 7969 served/68.2min/0 anomalies (daf30b04==origin/main) + Cloud clean 313/0-neterr + saturation 5000/0-neterr (INDEX sha-validated) + capacity baseline UID-matched
  • boundaries: overload degradation typed (0 sheds/121 drops/p99 10s — shed mechanism never observed firing), plane excludes ingest/webhook (Cloud C-004's), wave-1 r20/r50/rsat remain VOID (zero-served false greens re-typed by Collector and mechanized as zen-mgmt law capacity_evidence_served/v1)

4 receipt artifact(s) bound in the packet

DESTINATION_QUALIFICATION

GREEN (all switch preconditions on the destination side)

  • canonical transfer 5/5 digest-identical (c196-occ-001), dest Gate0 authority+idempotency GREEN, FORCE RLS verified + runtime-role parity, live anti-rollback 6/6 product-untouched, fence PASS 20/20

4 receipt artifact(s) bound in the packet

RESILIENCE

GREEN

  • 186 campaign restores 0 resurrections (today alone: 36 — rounds 12-17)
  • drills codified per durable-ops law as declarative Maestro workflow collector.drill-marathon@1.0.0 — first ONE-CALL execution GREEN (24 drills, 8 mid-write, 10.3s, receipt collector/drills/marathon-executions.json)
  • 25 journal-durable cycles

7 receipt artifact(s) bound in the packet

TENANCY_ISOLATION

GREEN

  • dest Gate0 authority clauses 403x5/own200/stale403 over transferred data
  • FORCE RLS + policy digest round-trip
  • runtime-role denial parity both sides

2 receipt artifact(s) bound in the packet

INDEPENDENT_VERIFICATION

CONSUMED — the C196 typed gap resolves

  • Support2 S210 adversarial night CLOSED 4/4 (shell outage resumed
  • battery 10/10
  • census zero-delta

2 receipt artifact(s) bound in the packet

ESTATE_CONTINUITY

GREEN (observing)

  • occurrence 46789292 re-verified live at 14:15Z during C198
  • Cloud C-011 cycles in flight (0 observed at 14:25Z)
  • observer window runs to ~18:00Z

3 receipt artifact(s) bound in the packet

The honest denominator

The DocsAI registry tracks 18 products × 47 evidence dimensions. Most cells are UNKNOWN — that is the true state of a young platform and we show it rather than hide it. Internal work-queue detail (which specific cells fail) stays off the public page.

Evidence grid — 2026-10-10

70 proven · 672 pending · 94 not applicable · 8 stale · 2 failing

report digest (merkle): 14da386e330bce6ffcbd32a…

Product denominator: advanced-cleaner, zen-cleaner, zen-control, zen-fabric, zen-flow, zen-identity, zen-lead, zen-lock, zen-maestro, zen-meerkat, zen-mgmt, zen-mon, zen-platform, zen-sdk, zen-support, zen-traffic, zen-trust, zen-watcher

Packet integrity

  • daedalus.bus-report — daedalus/reports/report sha256: cfb38f358649faea0e6f5ad4289cd0f9a3f9522e282d7d8ffbcc98df1cd5ba85
  • docsai.evidence-manifest-v1.1 — docs/static/ai/evidence/v1.1/manifest.json sha256: d7915f460a2c74f816d64ef1fff85e890f603837b86d8a02e317cb238062f88c · tranche C126
  • scout.r041-negatives — scout/reports/r041-negatives-receipt sha256: c2d6219350a75af6f03dd4976dfd13b190f81e2ebff91c2f1f9900f0d55df2ef · tranche R041
  • scout.r041-broker-seam — scout/reports/r041-broker-seam-finding sha256: c721cf9ffdb5609f607aa733045c82429d3fbf615b1b97d2bc51268345cb44b9 · tranche R041
  • mon.bus-report — mon/reports/report sha256: e1ffaa35dea65f6f8558554532f3bb6d977106329bcb0082cf6701fc0ba23988
  • docsai.products-registry — zen-mgmt/registries/products (20 product records) sha256: 22331e80f8216daedf6c7b2ff95fbe0830fff9911f36849496087c64a2d8e832
  • docsai.daily-quality — zen-mgmt/generated/quality/daily/2026-10-10.json sha256: d0bf80dcae28c37bbb9f6c642718c484d35831298f784c673a1c18bec6a31666
  • collector.go-nogo-v7 — collector/reports/go-nogo-decision-package-v7 sha256: f51a2119fbdb28ef033deee4ea23a3d22903d6f3779fe1b5ffd69da5ef0c0730 · tranche C198
  • platform.staging-roll-p135 — qual/website/platform-staging-roll-p135.json sha256: 49e8fb38151e38747a2764b14fdb141af3fc0a7b084664f6bccfab22595ca576 · tranche P135

All platform proof on this page is staging/sandbox/qualification-scoped. It is not a production-readiness, certification, or compliance claim. See the non-claims above and in the canonical manifest.

Current evidence baseline

Rendered from the generated public projection of the canonical docs evidence manifest (v1.1 manifest, source digest d83e6fc4131d…). Every status below is mechanically verified against that manifest.

  • L1 Foundation Freeze FROZEN — VICTORY LOCKED
    Evidence level: SANDBOX_CI · 4 explicit non-claims
  • L2 Connect Freeze PROVEN (SANDBOX)
    Evidence level: SANDBOX_CI · 6 explicit non-claims
  • Traffic V1.1 Engine Freeze FROZEN — VICTORY LOCKED
    Evidence level: SANDBOX_CI · 7 explicit non-claims
  • China Market Strategy IMPLEMENTATION PRESENT
    Evidence level: CODE_INSPECTION · 9 explicit non-claims
  • Payload Locality ARCHITECTURAL COMMITMENT
    Evidence level: CODE_INSPECTION · 2 explicit non-claims
  • Fabric Foundation Workload Enrollment (Mode A / Mode B) PROVEN (SANDBOX)
    Evidence level: SANDBOX_CI · 3 explicit non-claims
  • Canonical Connect Command/Query Convergence (REST / CLI / MCP) IMPLEMENTATION PRESENT
    Evidence level: CODE_INSPECTION · 3 explicit non-claims
  • Single Canonical Compiler & IR (ir/v1) ARCHITECTURAL COMMITMENT
    Evidence level: CODE_INSPECTION · 2 explicit non-claims
  • Trust / Lock Authority Separation & Bounded Custody ARCHITECTURAL COMMITMENT
    Evidence level: CODE_INSPECTION · 2 explicit non-claims
  • At-Least-Once Delivery Semantics ARCHITECTURAL COMMITMENT
    Evidence level: CODE_INSPECTION · 2 explicit non-claims
  • Domain-Specific Database Authority ARCHITECTURAL COMMITMENT
    Evidence level: CODE_INSPECTION · 1 explicit non-claim
  • Machine Enrollment Architecture ARCHITECTURAL COMMITMENT
    Evidence level: CODE_INSPECTION · 1 explicit non-claim

Frozen baselines

Frozen baselines are immutable qualification records; everything else is verified-current and expected to evolve. Frozen status is never inherited by upper layers.

  • L1 Foundation Freeze — counted repeated runs, all green, zero manual interventions (SANDBOX_CI) evidence_id: L1_FROZEN_GREEN · release binding: l1-foundation-frozen-green (DURABLE)
  • Traffic V1.1 Engine Freeze — counted repeated runs, all green, zero manual interventions (SANDBOX_CI) evidence_id: TRAFFIC_V11_ENGINE_FROZEN_GREEN · release binding: traffic-v11-engine-frozen-green (DURABLE)

Entry points

Machine-readable security

Narrative context (not proof)

Blog posts and the narrative registry explain why evidence matters. They are narrative_context only — not accepted_evidence, not proof_status, and not substitutes for manifest entries or hash-chain receipts.

Human guides

Proof status legend

FROZEN — VICTORY LOCKED
Immutable freeze artifact with counted repeated qualification runs; upper layers consume it as a contract.
PROVEN
Evidence artifact exists with stated scope (may still be local/mock or sandbox).
PARTIAL
Some criteria met; gaps documented in linked artifacts.
PLANNED
Not yet validated; roadmap or design only.
NOT_CLAIMED
Explicitly out of scope for the wedge or product surface.
BLOCKED
Cannot advance without owner decision or missing dependency.

Provider-by-provider signing status is documented in the provider documentation and the examples registry; this page no longer duplicates volatile per-provider statuses. Scope: local/sandbox header validation only; live E2E validation is cloud-gated.

Last updated: 2026-09-06 — from canonical evidence generated 2026-09-06T05:38:27Z. See non-claims and the V1 security validation summary for scope and limitations.

Currentness (OC137 D6 supersession note): the v1.1 evidence artifacts are frozen milestone projections — each carries the generated_at of its own validation set and is not refreshed in place. The live, packet-fed platform evidence projection on this page (regenerated from digest-verified fleet packets on every content build, generated 2026-10-14) is the current answer to "what is your current evidence"; where dates differ, the packet-fed projection is current. See the supersession note.

Third-party directory listings (e.g. llmstxt.org) are secondary — they may lag or fail indexing. Source of truth: the canonical docs evidence manifest and its stable channel; this page is a verified public projection of them. Internal methodology docs are not public certification.