> Neurobrick is an AI governance architecture that separates AI decision proposals from execution authority. Implemented in multi-asset financial research and execution environments. Historical findings, current scoped reconciliation, pinned local results and the evidence required for operating adoption and external reproduction.

<!-- neurobrick-document-version: evidence-first-v6; updated: 2026-10-22 -->

[neurobrick](<https://neurobrick.ai/>)

[principles](<https://neurobrick.ai/thesis>)

[architecture](<https://neurobrick.ai/architecture>)

[capabilities](<https://neurobrick.ai/capabilities>)

[authority](<https://neurobrick.ai/boundary>)

[observation](<https://neurobrick.ai/oversight>)

[evidence](<https://neurobrick.ai/evidence>)

[engineering](<https://neurobrick.ai/progress>)

[verification](<https://neurobrick.ai/verification>)

[scope](<https://neurobrick.ai/limits>)

Verification standard · dated observations

# A claim is only as strong as its weakest observation.

We improve the system by reproducing a wrong result, repairing its actual code path and testing whether a broken implementation is rejected. Every result keeps its source, population and limits.

Specific local behavior verified

Operating effect unverified

The unverified operating scope here belongs to the named local candidates. It does not mean trading or control implementation is absent. Other dated control observations and source selections are separate evidence families.

[Control observations and wider evidence index →](<https://neurobrick.ai/evidence#implementation-evidence>)

Evidence chain

## Observation → counterexample → repair → independent review.

1. 01

   Preserve the original failure
2. 02

   Exercise the actual code path
3. 03

   Reject bad implementations
4. 04

   Review the same source bytes

Publication, installed code and natural operating effect require their own observations. Passing this chain for one change does not certify the whole system.

Scoped engineering review

## A first failure must be visible.

A collector can fail during initialization, before creating its normal state file. Its failure was recorded in a separate heartbeat file, but the detector skipped that record. The local repair makes this first failure visible for current fleet members while preserving alert routing and duplicate suppression.

Specific local repair verified

Operating effect unverified

**Before repair**

5 of 15 cases failed

Includes two actual collector initialization failures missed by the detector.

**After repair**

15 of 15 cases passed

The same cases, with no errors or skips.

**Existing monitoring selection**

128 / 128 passed

All existing cases retained. This is a scoped selection, not a full-system result.

The tests call the actual collector, client constructor, state writer and detector in an isolated environment. They also cover a preserved domain cursor, one scan per daemon, cleared failures and excluded orphan heartbeats. This demonstrates a reachable code defect; it does not establish that it occurred in production.

### Risk decisions must use a committed verdict.

A separate Guardian candidate was checked against a pinned main tree. Its direct selection changed from 16 failures among 18 cases to 18 passing cases. The broader selection retained all 431 existing cases and passed 433 after the patch.

The 18 direct cases are included in the 433 broader cases. These populations are not added together. The candidate's required CI succeeded on its own exact head; compatibility and required CI for main 6ee6d730, observed in the 4 October 2026 at 22:24 KST record, were unverified in that record. The current reconciliation records later candidate publication checks separately.

### What this evidence still does not establish

The 4 October 2026 at 22:24 KST record did not verify the actual operating checkout and execution owner, loaded consumer revision, natural consumer completion, or rollback restoration with fresh output. A current operating census was not performed in the later public-claim audit. This local review does not establish public reproduction or external attestation.

[Dated engineering result and source pins →](<https://neurobrick.ai/evidence/2026-10-04-engineering.json>)

[SHA-256 →](<https://neurobrick.ai/evidence/2026-10-04-engineering.sha256>)

**Source, review and reproduction limits**

Separate internal read-only review checked the actual source and original test records. Four malformed evidence controls were rejected by the same reader. The core repository is private; published hashes identify internal bytes and are not an independent external attestation or public reproduction package.

stale\_daemon\_detector.run\_once\_sync · base 131d6b70fe3cd42d4c3f24cd910b609fb064136f
Repaired source SHA-256 18aacf166bef2721b26b30dc2fb12d1cf4d75a743be9e04596cbc7ed33ce14aa
Guardian candidate e9a241aca95ffa8713fad5d56b2fc6280a0ffe75 · tested main 83d3227d3cc0c491e37c9da46f702b7201396874

The earlier test-process probe below retains its own date, source and separate 18-case population.

Historical findings and current reconciliation

## A past finding is preserved. Its closure needs new evidence.

The review on 2026-10-10 separates original findings from later source and publication observations. UNKNOWN does not confirm a current defect or establish its closure. An implemented candidate with successful CI still needs its own operating verification.

Public claims, original-record integrity, current candidate PR metadata and one pinned source inspection; no runtime or live-state execution. Current session ownership was not established. System owners retain implementation and runtime verification; this registry records public evidence and leaves unobserved scopes open for verification.

[Reconciliation record and evidence IDs →](<https://neurobrick.ai/evidence/2026-10-10-claim-reconciliation.json>)

[Reconciliation file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-10-claim-reconciliation.sha256>)

Dated reconciliation on 2026-10-10: 34 finding families — 0 CLOSED\_VERIFIED; 27 UNKNOWN; 1 OPEN\_CONFIRMED; 5 IMPLEMENTED\_AWAITING\_VERIFICATION; 1 HISTORICAL\_ONLY. These counts describe the dated public evidence registry, not confirmed current defects or fleet-wide operating verification.

**Current status meanings**

- CLOSED\_VERIFIED: correction and verification are linked to the original finding for an exact named scope.
- IMPLEMENTED\_AWAITING\_VERIFICATION: implementation evidence exists; the claimed operating scope is not verified.
- REMEDIATION\_IN\_PROGRESS: active remediation ownership is confirmed; completion is not established.
- OPEN\_CONFIRMED: a new observation confirms the specific gap at its stated source or runtime scope.
- HISTORICAL\_ONLY: the statement describes a prior observation rather than the present state.
- STALE\_OR\_INACCURATE: a later scoped observation contradicts the current wording; the original remains preserved.
- UNKNOWN: evidence or access is insufficient to determine current state.

[Read the 34 finding families, original records and later observation scopes →](<https://neurobrick.ai/verification#public-claim-status>)

### NB-CF01 · End-to-end evidence linkage

**Historical observation · 2026-10-01**

None of the 772 broker-accepted stock orders carries a decision identifier; no accepted order on any venue is linked to a research signal (0 of 4,466; only stock-brokerage orders have a rule that can link one); and no index rule yet records the authority behind a decision, so a governed decision cannot be counted. The research-to-authorized-execution-to-accounting path is not established.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: research-to-authority-to-execution-to-accounting across engines. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-01-system-review.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-01-system-review.sha256>)

### NB-CF02 · Accepted-order decision identity

**Historical observation · 2026-10-01**

A daily lineage index (latest completed build 30 September) measures the path and a scheduled readiness run gates its first stage. None of the 772 broker-accepted stock orders carries a decision identifier yet, so no stock order links a decision to its accounting. A complete repeatable path is not yet established.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: historical accepted stock-order population. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-01-system-review.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-01-system-review.sha256>)

### NB-CF03 · Authority-event linkage

**Historical observation · 2026-10-01**

Give authority events their own identifiers so a governed decision can be counted, and read the release-authority artifact beside the approval axis.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: decision-to-authority index linkage. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-01-system-review.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-01-system-review.sha256>)

### NB-CF04 · DCA identity adoption

**Historical observation · 2026-09-30**

Merged and installed source contains identity minting. Observed live rows still lack decision IDs; runtime adoption is unresolved.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: DCA producer and natural consumer adoption. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress.sha256>)

### NB-CF05 · Research readiness input failure

**Historical observation · 2026-10-01**

The latest research readiness run stopped at its data-panel lineage check with 17 of 42 factor checks failing, so no research observation started.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: one dated research readiness run. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-01-system-review.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-01-system-review.sha256>)

### NB-CF06 · Unread release-authority artifact

**Historical observation · 2026-10-01**

The release-authority artifact was not read; this review makes no claim about central live release.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: release-authority observation access. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-01-system-review.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-01-system-review.sha256>)

### NB-CF07 · Approval actor and grant time

**Historical observation · 2026-10-01**

HTTP availability is not business readiness. The approval-axis value is a summary the health endpoint derives from the control plane's approval settings; no order is admitted or refused by reading that summary. The kill switch is reported separately. When the approval was last granted was not observed.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: approval-axis dated observation. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-01-system-review.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-01-system-review.sha256>)

### NB-CF08 · Server-side risk refusal coverage

**Historical observation · 2026-09-27**

Independent execution engines retain their own permissions, account contexts and gates. On the primary engine the core book remained sealed: identified order attempts on 26 September (KST) were refused by the server-side risk check and the core sell block.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: specific refused execution context. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-27-system-review.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-27-system-review.sha256>)

### NB-CF09 · Autonomous submission conditions

**Historical observation · 2026-09-27**

The operator approval that the live-execution safety gate requires is granted, so the control plane reports its approval axis as open. The two autonomous-release opt-ins are off, and in the inspected half hour every autonomous submission attempt was skipped by a further gate (opt-in contract not active, admission lease not ready, certified trial not active). The portfolio kill switch is ARMED.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: dated opt-in, lease and trial progression. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-27-system-review.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-27-system-review.sha256>)

### NB-CF10 · Portfolio kill configuration versus trigger

**Historical observation · 2026-10-01**

ARMED

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: reported kill configuration and natural trigger. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-01-system-review.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-01-system-review.sha256>)

### NB-CF11 · Kill/release event identity

**Historical observation · 2026-09-30**

Merged and installed source contains event identity. The inspected live ledger was empty; no event was triggered for this review.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: trigger/clear event identity and natural execution. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress.sha256>)

### NB-CF12 · Portfolio target authority consumption

**Historical observation · 2026-09-30**

Merged and installed source contains target identity. Runtime consumption and authority linkage remain unverified.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: target producer and authority consumer. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress.sha256>)

### NB-CF13 · Read-only HTTP deployment

**Historical observation · 2026-09-30**

Merged source tightens signal history and raw linkage boundaries. Deployed HTTP behavior was not independently checked.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: deployed console HTTP behavior. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress.sha256>)

### NB-CF14 · Disabled inventory issuance

**Historical observation · 2026-09-30**

Merged implementation remains disabled by default. Runtime manifest issuance and adoption remain unverified.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: manifest issuance and runtime adoption. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress.sha256>)

### NB-CF15 · Descendant-process cleanup

**Historical observation · 2026-10-02**

Controlled dependencies exercise the termination branches. Actual descendant-process reaping is not established.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: actual descendant-process reaping. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-02-verification.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-02-verification.sha256>)

### NB-CF16 · Absolute OS wall-clock bound

**Historical observation · 2026-10-02**

This bounds the tested cleanup read. It does not prove an absolute wall-clock bound for every operating-system interaction.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: all operating-system interactions. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-02-verification.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-02-verification.sha256>)

### NB-CF17 · Initialization first-failure visibility

**Historical observation · 2026-10-04**

local\_candidate

**Current status · checked 2026-10-10**

OPEN\_CONFIRMED

The state-file selection omission is still present in the newly inspected source revision. The dated local repair remains preserved. This confirms a source gap only; operating occurrence and installation are UNKNOWN.

Scope: stale\_daemon\_detector.run\_once\_sync source before a domain state exists. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-04-engineering.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-04-engineering.sha256>)

[Later observation · NB-SOURCE-20261010-INIT →](<https://neurobrick.ai/evidence/2026-10-10-claim-reconciliation.json#/records/0>)

### NB-CF18 · Guardian main compatibility

**Historical observation · 2026-10-04**

unverified

**Current status · checked 2026-10-10**

IMPLEMENTED\_AWAITING\_VERIFICATION

The dated implementation candidate is still open and unmerged. Named exact-head lint, regression and test checks succeeded at this review. Current installation, natural execution and durable operating effect were not verified.

Scope: candidate versus the separately recorded main revision. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-04-engineering.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-04-engineering.sha256>)

[Later observation · NB-PR-20261010-5084 →](<https://neurobrick.ai/evidence/2026-10-10-claim-reconciliation.json#/records/5>)

### NB-CF19 · Operating adoption and durable rollback

**Historical observation · 2026-10-04**

loaded\_consumer\_revision

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: installed revision, execution owner, natural completion and rollback. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-04-engineering.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-04-engineering.sha256>)

### NB-CF20 · Research input measurement boundary

**Historical observation · 2026-09-30**

All absent inputs remain missing; partial invalid inputs become unknown. Completeness covers only listed/discovered files, and older reports acquire no invented unknown count. Freshness, PIT correctness and research quality are not certified.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: input timestamp measurement, not research-quality certification. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress-round2.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress-round2.sha256>)

### NB-CF21 · UNKNOWN freshness denominator

**Historical observation · 2026-09-30**

Only evaluable states enter the rate. Unknown, unparsed and unprobed rows remain visible and excluded from both sides; no evaluable observations yield 0/0 with no rate.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: evaluable freshness states and observer coverage. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress-round2.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress-round2.sha256>)

### NB-CF22 · Native and missing producer identities

**Historical observation · 2026-09-30**

Direct nonstring native decision references are rejected. Normal opaque IDs, multiple children and content identities remain compatible. Already stringified producer IDs cannot reveal their original type; missing live DCA identities remain unresolved.

**Current status · checked 2026-10-10**

IMPLEMENTED\_AWAITING\_VERIFICATION

The dated implementation candidate is still open and unmerged. Named exact-head lint, regression and test checks succeeded at this review. Current installation, natural execution and durable operating effect were not verified.

Scope: producer identity type and live DCA identities. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress-round2.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress-round2.sha256>)

[Later observation · NB-PR-20261010-4923 →](<https://neurobrick.ai/evidence/2026-10-10-claim-reconciliation.json#/records/2>)

### NB-CF23 · Heartbeat health interpretation

**Historical observation · 2026-09-30**

Existing staleness and classification priorities remain. PID presence and a parsed timestamp do not establish installed code identity or successful application work.

**Current status · checked 2026-10-10**

IMPLEMENTED\_AWAITING\_VERIFICATION

The dated implementation candidate is still open and unmerged. Named exact-head lint, regression and test checks succeeded at this review. Current installation, natural execution and durable operating effect were not verified.

Scope: process heartbeat versus completed application work. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-operational-proof-publication.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-operational-proof-publication.sha256>)

[Later observation · NB-PR-20261010-4929 →](<https://neurobrick.ai/evidence/2026-10-10-claim-reconciliation.json#/records/3>)

### NB-CF24 · Carried observation provenance

**Historical observation · 2026-09-30**

Partial unreadable input follows the existing degraded path. Carried values retain degraded status, carry markers and their original observation time across repeated publication; legacy records acquire no fabricated provenance.

**Current status · checked 2026-10-10**

IMPLEMENTED\_AWAITING\_VERIFICATION

The dated implementation candidate is still open and unmerged. Named exact-head lint, regression and test checks succeeded at this review. Current installation, natural execution and durable operating effect were not verified.

Scope: carried/degraded status and original observation time. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-operational-proof-publication.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-operational-proof-publication.sha256>)

[Later observation · NB-PR-20261010-4916 →](<https://neurobrick.ai/evidence/2026-10-10-claim-reconciliation.json#/records/1>)

### NB-CF25 · Recovery domain completion

**Historical observation · 2026-09-30**

HEALTHY remains process/log health, not proof of completed domain work or durable restoration. Naive health timestamps, malformed target evidence and legacy absence remain unknown.

**Current status · checked 2026-10-10**

IMPLEMENTED\_AWAITING\_VERIFICATION

The dated implementation candidate is still open and unmerged. Named exact-head lint, regression and test checks succeeded at this review. Current installation, natural execution and durable operating effect were not verified.

Scope: process/log health versus durable domain restoration. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-operational-proof-publication.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-operational-proof-publication.sha256>)

[Later observation · NB-PR-20261010-4932 →](<https://neurobrick.ai/evidence/2026-10-10-claim-reconciliation.json#/records/4>)

### NB-CF26 · Observer independence and coverage

**Historical observation · 2026-10-01**

Observer independence and complete coverage were not re-attested. Scheduled-task success alone does not establish either.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: observer independence and full coverage census. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-01-system-review.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-01-system-review.sha256>)

### NB-CF27 · Fleet clean-box recovery

**Historical observation · 2026-10-01**

NAS mounts and running services do not establish a verified fleet-wide clean-box recovery or a reconciled formal close.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: fleet-wide restore and formal close. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-01-system-review.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-01-system-review.sha256>)

### NB-CF28 · Historical transport failures

**Historical observation · 2026-09-30**

Name-resolution and host-key verification failures left two observation paths unavailable. Host health and external attestation were not established.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: two dated unavailable observation paths. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress.sha256>)

### NB-CF29 · Engine-specific market integration boundary

**Public statement captured · 2026-10-10**

Generic KIS derivative scaffold branches are sealed. Source integrations do not establish every product is currently enabled.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: engine-specific adapter paths and current product enablement. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/architecture#market-coverage>)

### NB-CF30 · AVA server and inference verification

**Public statement captured · 2026-10-10**

AVA server implementation, model residency and current distributed inference were not verified by this source review.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: AVA server implementation, residency and distributed inference. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/architecture#market-coverage>)

### NB-CF31 · Historical CI and publication stage

**Historical observation · 2026-09-30**

Reviewed Draft \#4916 updated; exact-head CI was queued or running at this review.

**Current status · checked 2026-10-10**

HISTORICAL\_ONLY

The queued/running statement describes 30 September. On 10 October, named exact-head checks for NeuroPublic \#4916, \#4923, \#4929, \#4932 and \#5084 had completed successfully, while all five PRs remained open, Draft and unmerged. The fleet-brain \#1500 latest state was not inspected.

Scope: candidate publication and named exact-head CI only. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress-round2.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-09-30-engineering-progress-round2.sha256>)

[Later observation · NB-PR-20261010-4916 →](<https://neurobrick.ai/evidence/2026-10-10-claim-reconciliation.json#/records/1>)

### NB-CF32 · Historical stale ledgers

**Historical observation · 2026-08-30**

One domestic order ledger has no declared freshness ceiling.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: dated domestic ledger and rotation freshness. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-08-30-current.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-08-30-current.sha256>)

### NB-CF33 · Replay and effect remeasurement

**Historical observation · 2026-10-01**

Authority-surface attribution, replay equivalence and remediation-effect totals were not re-measured. August figures remain historical evidence.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: authority attribution, replay and effect totals. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-01-system-review.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-01-system-review.sha256>)

### NB-CF34 · External reproduction and attestation

**Historical observation · 2026-10-02**

The local result is tied to private source and test bytes. Public JSON and SHA-256 identify this disclosure; they do not independently reproduce or attest to the private execution. The website claim checks can be run from the website repository.

**Current status · checked 2026-10-10**

UNKNOWN

No new observation in this audit establishes the current operating state for this claim. The dated finding remains available; UNKNOWN is neither closure nor confirmation of a present defect.

Scope: independent reproduction of private execution. An identified system owner must provide a commit-pinned observation for this exact scope before operating closure can be published.

[Original finding →](<https://neurobrick.ai/evidence/2026-10-02-verification.json>)

[Original file SHA-256 →](<https://neurobrick.ai/evidence/2026-10-02-verification.sha256>)

Original finding → implementation evidence → scoped re-verification → current status. The source finding confirms a code path at a pinned revision; it does not establish an operating incident. The 5 candidate PR observations establish named CI and publication states. They do not establish deployment, natural consumer completion, rollback or external attestation.

Earlier scoped implementation

## The test observer must preserve uncertainty.

The earlier local repair strengthens failure cleanup and output interpretation in the test probe. Unavailable captured output remains unverified, even when the process reports a successful exit.

**Local test population**

18 / 18

0 failed · 0 errors · 0 skipped

**Bad implementations rejected**

6

Controls check that verification can fail.

**Review level**

Separate internal review

External attestation remains unverified.

### Cleanup failure stays unresolved

**Before**

An unsuccessful Windows tree-termination command could return without attempting the direct-child fallback.

**After**

A nonzero tree-termination result attempts the direct-child fallback. Cleanup failure remains an unknown outcome.

Controlled dependencies exercise the termination branches. Actual descendant-process reaping is not established.

### Timeout cleanup has a finite observation window

**Before**

An implicit process context exit or an unbounded output read could keep the probe waiting after the execution deadline.

**After**

The probe retains its execution deadline and limits the post-failure output read to five seconds. Unresolved cleanup returns a timeout outcome.

This bounds the tested cleanup read. It does not prove an absolute wall-clock bound for every operating-system interaction.

### Unreadable output cannot become a pass

**Before**

Locale-dependent text capture could lose UTF-8 output or interrupt attribution; unavailable output could also fail attribution.

**After**

Binary capture is decoded strictly as UTF-8. Unavailable captured output or a UTF-8 decoding failure produces an explicit unverified outcome even with a zero exit code.

One isolated real child exercises UTF-8 output across a forced legacy text-decoding boundary. The other cleanup counterexamples use controlled process dependencies. This verifies capture availability and decoding, not output completeness; valid empty output retains the process exit-code result.

[Dated result and source pins →](<https://neurobrick.ai/evidence/2026-10-02-verification.json>)

[SHA-256 →](<https://neurobrick.ai/evidence/2026-10-02-verification.sha256>)

[Earlier improvements →](<https://neurobrick.ai/progress>)

**Reproduction and access limits**

The local result is tied to private source and test bytes. Public JSON and SHA-256 identify this disclosure; they do not independently reproduce or attest to the private execution. The website claim checks can be run from the website repository.

mxroach/neuropublic · kill\_tree / run\_pytest\_once
Commit d6e06a514f5bf07ad7d1d3205975c87a9d611cd9
Source SHA-256 75f6296470174813368f0c95b0bee13b2d15eec4f7c474a8de9fff0bfd15cf91

Earlier disclosures retain their own dates, populations and publication stages. This local test result does not refresh fleet observations or resolve the held full-regression candidate.

Acceptance criteria

## What would justify a stronger claim.

### Operating effect

Not verified

The installed revision, a natural-run observation from the affected component, declared coverage and a comparison window.

### End-to-end decision evidence

Not verified

Research, authorization, execution and accounting records sharing consistent identifiers; missing records retained in the denominators.

### External reproducibility

Not verified

A sanitized runnable reproduction bundle and an observer able to check the original evidence independently.

These criteria were not verified in the 2 October 2026 at 20:45 KST record. The reconciliation above distinguishes that finding from later scoped observations; it is not a current runtime census. This page covers technical claims. Internal review and public hashes do not constitute external attestation.

Neurobrick · systems, authority and evidence

Last updated · 22 October 2026

[Public scope →](<https://neurobrick.ai/limits>)

[Canonical page](<https://neurobrick.ai/verification>)
