Learning portalDirectory audit: 17 November 2026
This is a bounded inspection, not a guarantee that every external service will remain available.
Internal structure
176 HTML pages and 2,640 local references scanned, including the complete encyclopedia. No missing anchors or duplicate HTML IDs detected. Relative paths beginning ./ were checked against their normalized file names.
Confirmed repair
The 200-lesson directory spans two pages. Filters previously hid matching lessons held on the other page. Both pages now list matching lessons on the other page with direct lesson anchors. All 200 external resource labels now state that they open a resource collection. Added guidance explains country coverage, eligibility, fees and the difference between self-reported progress and accreditation.
Live interaction checks
- Course search: matching result, no-result state, department filter and cross-page lesson links.
- 34 lesson disclosure controls opened during this audit.
- Opportunity directory: Canada and Employment filters; actual Job Bank link click reached the Canadian service.
- Expanded Academy: Biology filter and course selection.
- Language directory: Turkish search and next-page control.
- Employment directory: trucking occupation search returned occupational matches.
External availability screen
767 distinct external addresses were submitted for retrieval. 271 returned readable content; 496 returned retrieval errors or unavailable content in this environment. These are unresolved results, not proof that those websites are broken for visitors. Retrieval is separate from a live browser click and from verifying every factual statement. Dynamic links generated by scripts require additional inspection.
Remaining verification
A complete click-by-click assessment of every external destination, dynamic result, language and form is not yet established. Encyclopedia interaction testing encountered a browser timeout. External destinations can impose sign-in, regional restrictions or technical limitations. No resource was removed solely because this audit environment could not retrieve it.
Open the repaired course directory · Opportunity directory
Live browser retest — 18 November 2026 UTC / 17 November Pacific
Actual click tests opened all 200 of 200 introductory lesson disclosure controls successfully, across both directory pages. This is complete coverage of those disclosure controls, not every site interaction or every external source.
Confirmed defect repaired and retested: cross-page search status previously displayed only the current page count and could report no matches while the other page held results. The shared search enhancement now reports total matches out of 200, current-page and other-page counts, and accurate empty-state guidance. Cybersecurity returned 10 matches on the other page; a nonexistent term returned zero across both pages; an income search on page two returned Lesson 1, and clicking opened that lesson on page one.
Additional live checks: Library English selection, zero-result search and clearing; Research science search (22 results, 12 displayed), zero-result search and clearing; encyclopedia biography disclosure opened and Douglass search returned Frederick Douglass; main records filter accepted input and restored results after clearing. National Academies link actually opened Decoding Science with its scientific-method content. The initially unresponsive biography input was inside a collapsed disclosure and worked after opening it; no site defect was established there.
One transient upstream connection termination recovered on a single reload. It remains an availability observation, not a reproducible application defect. Browser logs contained extension metadata errors and a redacted blob error; these are not evidence of a clean site-wide runtime audit and require attribution if repeated.
Outstanding: exhaustive external destination clicks, complete forms/download/language/mobile testing and all other site controls. Hourly monitoring remains enabled; no 100% site-wide acceptance is claimed.
Deep verification pass — 17 November Pacific / 18 November UTC
Complete source integrity scan
Current inventory: 277 files, including 186 HTML files. All 2,664 local href/src references were normalized and checked against deployed files and HTML anchor targets. No missing files, missing anchors, duplicate HTML IDs, unsafe javascript/vbscript links or duplicate external script tags were detected. One apparent empty href was JavaScript selector construction in the verification utility, not an empty rendered hyperlink. This is source integrity evidence, not complete browser acceptance.
Confirmed defect and repair
Financial Academy cards 24–33 still said planned and had no course link despite all ten corresponding course files being published. Updated each card to link directly to its course and corrected the availability summary. Retained pending functional, translation and independent-review qualifications. Actual accessibility-browser clicks then tested all 33 course links from the catalogue; all 33 reached their expected path and displayed the correct course heading. Earlier locator-only navigation attempts did not establish navigation; the recorded pass counts use subsequent observed accessibility clicks. These link passes do not certify the full course workflows.
Additional actual browser tests
- Tools: default cash-flow calculation 2,500 minus 2,140 = 360; break-even ceil(600/27) = 23 units; equal unit price and cost produces no finite break-even; fictional action-plan generation includes entered action, barrier and source. Cleared tab contents by reload.
- Resume builder: empty required name blocks submission; fictional name/role/summary produces matching preview; text-download event observed; clear confirmation accepted and entries/preview cleared. Download content and print-PDF fidelity remain unverified.
- Application tracker: empty employer/title validation; fictional entry addition; Interview status selection; backup-download event; entry removal; zero entries after reload. Backup import remains untested.
- Job-offer safety review: no-selection result explicitly does not verify an offer; selecting the upfront-payment warning returns one warning and relevant steps; reset clears selections.
- Course 33 capstone: default calculator returned monthly surplus 550 CU, runway 0.92 months, two-month shock balance 200 CU; empty assessment returned 0/10; completion-record button remained disabled; retake and course reset accepted. Test-only course state reset. Full passing assessment/certificate workflow remains untested.
Open acceptance scope
Full-site completion is not established. Remaining work includes all untested page controls, all course passing/failing assessment and persistence paths, download-file contents, backup restore, certificate rendering, supported-language fidelity, mobile/keyboard review and unresolved external destination clicks. Hourly audit should resume from these outstanding items and preserve the repaired ten course links.
Download, language and responsive verification — 18 November Pacific
This is additional bounded evidence; full-site acceptance remains incomplete. Structured verification checkpoint.
Certificate and notice files
Added direct local PDF downloads and separate notices to eight signing editions: Spanish, French, Portuguese, Arabic, Hindi, Indonesian, Bengali and Urdu. All 16 resulting files were downloaded, parsed as one-page PDFs without repair warnings, and visually inspected for native text, date, signature and clipping. Eight signature/consent gate tests passed. English and Turkish signed certificates and the separate English notice were also parsed as valid one-page files. The blank oath is one page; the legacy signature form is two pages with its signature widget on page two. A 90-character signature wrapped inside the certificate frame.
Limits: multilingual PDFs are image-based; searchable text, PDF/UA and physical Android/iOS operation are not established.
Language controls and responsive layouts
Localized shared headings, notes controls and save messages in four global learning modules. All 52 module/language selections (four modules × 13 languages) passed language, direction and control-label checks. Turkish notes persisted after reload in all four modules. Expanded English-only material is identified; independent linguistic review and complete site translation remain outstanding.
Rendered all 185 visitor pages at 340 CSS pixels using desktop Chromium at 400% zoom. No horizontal page overflow was observed; 18 transient load failures passed one targeted retry each. Corrected two skip links that overlapped navigation; keyboard focus and activation were retested. This does not establish physical-phone or exhaustive touch acceptance.
Actual external clicks
Clicked all 33 primary opportunity-directory destinations: 26 returned recognizable content; seven remained blocked, partial or technically unresolved (CareerOneStop, Skill India, Services Australia, South Africa Labour, SEDA, Philippines DOLE and Philippines DTI). Corrected UNICEF to a relevant child online-safety article and Europass to its English page; both final destinations were re-clicked and observed. External restrictions were not bypassed. Other bibliography destinations remain outstanding.
Capstone completion-record workflow
Course 33: empty assessment scored 0/10 and kept download disabled; correct answers scored 10/10. Five stage selections, a clearly marked QA draft, self-attestation and the best score survived reload. The actual downloaded SVG parsed as valid XML (1000 × 650) and contained the correct course, 100% local score, UTC generation date, self-assessment and non-accreditation limitations. Revoking self-attestation disabled the download. Disposable test state was reset and the empty draft, zero stages and zero best score were observed afterward. This was a technical QA test, not genuine learner completion. The other 32 complete course workflows remain unverified.
Still required
Remaining course assessments and records, tracker restore and other download contents, full translation fidelity, physical mobile interactions, remaining external destinations and all other untested controls and integrations. Do not interpret these results as 100% acceptance.
Course-record verification and repairs — 18 November Pacific
Completion-record file contents: 33 of 33 courses checked. This is file-content coverage, not complete course or whole-site acceptance. Detailed checkpoint and limitations.
Downloaded 32 distinct course records during this pass, supplementing the previously checked Course 33 file. All parsed as valid SVG XML with the expected course, date and completion limitations. Rendered the 32-file set for visual inspection. Course 13's inherited centering was ignored by one renderer; added explicit text centering to each line, regenerated the actual download and confirmed no text overflow.
Reproduced and repaired Course 25's lost written-project checkbox: it had not been saved, causing eligibility to disappear after reload. The final browser retest retained stages, attestation and certificate eligibility after reload, downloaded the record and cleared disposable test progress.
Added course-specific reset controls with confirmation on 29 pages lacking one; preserved original controls on four pages. Removed a redundant addition from Course 24 after its original reset was tested.
Detailed new empty/passing assessment, reload and download tests covered Courses 02–05, 13–19, 21–32, with test cleanup. These checks do not cover every threshold or failure scenario. Courses 01, 08 and 12 produced files before a tool-session timeout; their individual feedback output was not retained, so those states need an evidence-preserving retest. Course 12 test state was explicitly cleared after recovery.
Preserved pre-existing progress in Courses 06, 07, 09, 10, 11 and 20. For 06, 07 and 09, temporarily checking the missing written-project control unlocked the record and persisted after reload; restored that control to unchecked and observed the original five stages, 100% best score and locked record. Courses 10, 11 and 20 downloads were checked without replaying their complete assessments.
Remaining: course boundary/failure scenarios and remaining assessment replays; all other untested controls/integrations; translation fidelity; physical mobile interaction; and unresolved external sources. Do not mark the entire site 100% verified.