What the evidence refers to
J=0 / Jameson Zero Condition is the broader conceptual reference problem. Stasrift is an engineering application. JC-001 is a specific adversarial evidence test. A JC-001 result is scoped to its procedure and environment, not proof of the broader concept or every engineering application.
Read the distinctions. Dated reports below remain historical evidence for the releases they name.
September 29 · Browser local node prototype
DESIGNED: a portable node model with a replaceable local persistence adapter. Device and node identity remain separate from owner identity and authority.
IMPLEMENTED: stable logical Node ID, versioned JSON format, atomic migration of the existing IndexedDB record, snapshot and chain verification, open export and validated import into unused storage. Deletion removes payloads and old version fingerprints, retaining a minimal receipt.
TESTED: 28 automated tests passed. LN-001 and LN-002 passed for actual tab closure and reopen. LN-003 through LN-006 passed within the documented local browser scope, including stored tampering, actual download and import into independent origin storage, and deletion without removed content in the new export. JC-001 passed a fresh execution with a child exit of 23 and a denied evidence rewrite.
NOT YET PROVEN: full browser process restart, power loss durability, physical erasure, trusted authorship, resistance to host compromise or multi node reconciliation. Browser process launch was blocked in the test environment. This is a browser node layer, not a production grade distributed node. Hashes on the same host can be rewritten together.
Inspected architecture and open format · Raw automated results · Browser observations and exact scope · Fresh JC-001 evidence · Failures and corrections · Reproduce these checks
September 29 · Home and verification refinement
At this September 29 revision, Home led with Preserve · Ask · Verify. The October 6 company update moves that prototype below the company and ecosystem introduction. The local text comparison identifies the changed span, including equal-length edits. Show the Record links directly to saved versions and a visible verification result.
JC-001 v3 ran a real operation that exited 23 while the agent fixture claimed completion. A scripted attempt to rewrite the observer’s evidence was denied, and before/after evidence hashes matched. Result: CLAIM CONFLICTS WITH RECORDED EVIDENCE.
This tests one denied filesystem write under Node permissions. It is not an OS sandbox and does not prove containment of arbitrary malicious code. The observer and host are trusted; the browser prototype does not inherit this permission boundary.
Verification and limitations · Raw regression results · Show the Record: JC-001 v3 · Reproduce the test · Failures and corrections · First regression attempt
Local record loop · September 29 revision
Implemented: transactional browser-local text versions, sequence fingerprints, a deterministic change summary, Show the Record, export, and user-selected content deletion with a minimal receipt. The existing fixed recovery demonstration remains separate.
Current verification and limits · Current raw test output · JC-001 claim and independent execution observations · Reproduction harness · Comparison and verification source
JC-001 uses an intentionally false agent fixture; no live AI is claimed. The observer is separate from the child process but runs on the same machine. A compromised observer or host remains outside this test’s guarantees.
DESIGNED
The Continuity Framework defines intended behavior and conditions for recovery. The public Field is a deterministic teaching example, not production telemetry or a durability test. Reproduce the current JC-001 with the PowerShell harness, the failing child, and the evaluator in the source package. The blocked first attempt and failure notes remain available.
IMPLEMENTED
The public site includes local assets, an allowlisted offline cache, and an interactive demonstration with browser storage. The demonstration reuses the existing topology and Rebound model. It is separate from the authenticated record system.
TESTED
Public website and demonstration · local review v0.3
The archived September 28 local review recorded 16 passing automated tests. They check the five-section Home and navigation, and exercise the teaching model, local save failure handling, event history reconstruction, routing, security headers, and an offline cache fixture. Local browser checks covered keyboard activation through the conflict and recovery sequence, retained history after reload, and responsive layouts. Detailed outcomes and untested areas are recorded in the local review report.
Read versioned demonstration procedures, actual results, and limitations · Raw test output
These results do not prove production durability or complete Continuity recovery. No independent accessibility audit or physical touchscreen trial is claimed.
Existing integration tests
An archived September 28, 2026 report records nine successful foundation integration tests against an Arkosom v4.2.0 source package using local fixtures. These tests were not rerun during this public-site editorial review. They exercised existing Arkosom and Data Line code, including lost acknowledgement retry, duplicate prevention, retrieval verification, tamper rejection, context isolation, retention expiry, and session revocation.
These results apply to that package and fixture environment. They do not establish current production health or complete Continuity recovery. Read the scoped test record.
NOT YET PROVEN
A full end to end Continuity report is not yet published. Browser restart durability, storage failure, multi node conflict reconciliation, required redundancy, and verified Rebound need explicit procedures and recorded outcomes. No successful outcome is assumed.
Versioned Continuity tests
CT-001 v0.1: Local commit survives network removal and restart. Status: NOT EXECUTED as a complete production Continuity test. Environment and actual result: not yet recorded. The template below defines procedure, expected result, actual result, and limitations for each test.
Reproducible report structure
Each report should identify the source revision or package hashes, environment, storage and retention assumptions, failure injection, expected result, actual observations, raw evidence, limitations, and reviewer. Publish corrections as later versions rather than silently changing a result.
Download the unexecuted test report template
- Create Record A while connected and verify the local commit and receipt.
- Disable networking. Restart offline and retrieve A.
- Create B and C offline. Restart again and retrieve all three.
- Restore connectivity. Verify acknowledgements, retrieval, integrity, and duplicate prevention.
- Create conflicting changes on isolated nodes. Reconnect and verify quarantine and visible conflict without silent overwrite.
- Resolve the conflict through the required authority, reconcile history, and verify required paths and redundancy.
- Evaluate every Rebound condition and retain the interruption and reconciliation history under the applicable retention policy.
Results must say what was tested and what was not tested. A passing local test cannot establish universal fault tolerance or an accessibility conformance claim.