MapleJockey demonstrator verification
12 November 2026. Local founder-led increment; no new external spend, public deployment, production modification or outreach.
Implemented coverage
The self-contained HTML prototype provides workload/service inventory, dependency/data-flow views, evidence records, claim provenance and freshness/conflict states, requirements and linked findings, prioritised actions, local edits, browser save/load, structured JSON export/import, claims CSV and a readable Markdown report. New assessment and scope editing support a fresh pilot record.
The internal case contains eight service/component nodes, eight flows, twelve evidence records, fourteen claims, seven requirements, three findings and three actions. Five claims are Unknown. A separate invented fixture demonstrates all seven required evidence labels. No real supplier conflict or stale event is asserted.
Automated verification
- 10 prototype tests cover the real fixtures, all seven labels, full structured record round trips, import-origin warnings, unknown/stale findings, invalid links/dates, edit history, blank pilots, report completeness and standalone network-denying packaging.
- Independent review tests cover duplicate IDs, required fields, reference integrity, evidence states, indirect requirement uncertainty, invalid dates, escaping, CSV injection defence, immutable edit history and import provenance. See the packaged independent review and exact result count.
- The isolated Internal reporting application passed 12 functional tests and syntax checking. These execute the actual restored model, not a mocked archive check.
Browser checks completed
Verified overview rendering, claim filtering, claim-to-evidence drill-down, action editing, explicit browser save, reload and load, and separation of the illustrative conflict case. The test action was restored by clearing the disposable local copy and reloading the bundled case.
Verified the Internal reporting application rendered its synthetic dataset, changed the reporting window, accepted a synthetic comparison and downloaded a JSON export. The subsequent file chooser stalled, and the browser session later timed out. Browser JSON re-import is therefore unverified. New assessment and Edit scope were added after that browser session; source syntax and core validation are checked, but those final two UI controls have not been browser-verified.
Recovery result
Target: restore a sanitised static application and generated report bundle into a fresh local directory and run usable model operations within 60 seconds. This is a demonstrator threshold, not a business-approved RTO.
Nine of nine restored file hashes matched. Archive creation took 0.003532 seconds, extraction 0.002351 seconds and functional checks 0.404820 seconds, for 0.410703 seconds total. The model source was unchanged. The UI was sanitised to label fixtures, isolate browser storage and remove external navigation. The original report bundle was not read or copied. Original source remained clean and unchanged.
The result supports this local code/model slice plus the separately observed browser functions. It does not establish recovery of production data, all browser-created state, platform authentication, DNS/CDN/TLS, supplier backups, keys or production throughput. The source-level manual-comparison import omission remains a P1 finding. No claim of full production recovery is made.
Boundaries
No encrypted evidence vault, authentication/authorisation service, cryptographic attestation, connectors, legal opinion or certification is implemented. Imported records and their histories are assertions, not authenticated evidence. Local files and browser profiles need user-managed protection. Responsive styles are present, but a mobile-device check was not completed. External provider facts remain Unknown pending Nine questions for missing evidence.
Evidence from the demonstration. Independent assessment checks · Recovery measurements · Browser observations.