Evidence edition / 23 October 2026 / en-US
Coverage with a denominator
25 executed of 31 bounded cases: 22 passed, 1 failed, 1 partial, 1 observation; 1 conditional check not offered, 3 blocked, 2 planned/unexecuted. An executed partial check is not a pass. These counts are not a percentage of the Xbox product.
Discovery
1 executed / 1 bounded cases
CX-01
Search
3 executed / 3 bounded cases
CX-02, CX-03, CX-06
Details
2 executed / 2 bounded cases
CX-04, CX-14
Navigation
3 executed / 3 bounded cases
CX-05, CX-18, CX-25
Recovery
2 executed / 2 bounded cases
CX-07, CX-19
Access
3 executed / 3 bounded cases
CX-08, CX-15, CX-26
Keyboard
3 executed / 3 bounded cases
CX-09, CX-10, CX-24
Browse
2 executed / 3 bounded cases
CX-11, CX-12, CX-13
Responsive
6 executed / 6 bounded cases
CX-16, CX-17, CX-20, CX-21, CX-22, CX-23
Accessibility
0 executed / 2 bounded cases
CX-27, CX-30
Performance
0 executed / 1 bounded cases
CX-28
Compatibility
0 executed / 2 bounded cases
CX-29, CX-31
Route population
Seven distinct public route identities: home, search, All games, Family Friendly, Fortnite, 1000xRESIST and ASTRONEER. Query changes are states of search, not new routes. Two collection routes and three details are representative samples, not an exhaustive catalog.
| Route family | Sample | Uncovered |
|---|---|---|
| Home | Desktop, 390px, 320px, plan modal, guide | Every carousel item and footer destination |
| Search | Fortnite, unmatched query, clear, Back | Typos/ranking breadth, locale equivalence |
| Collections | All games; Family Friendly | Other genres, entire catalog, full infinite scrolling |
| Details | Fortnite; 1000xRESIST; ASTRONEER | Every language/capability/gallery control; age-gated media |
CARBON planning population
The installed runtime generated 110 broad candidates. They remain candidates. A separate reviewer proposed 28 product and 2 report-integrity candidates. The 31-case execution ledger is the curated assessment record; these pools must not be added together.
Confidence by quality concern
Functional, usability, accessibility and responsive evidence is bounded and partial. Performance, security, privacy internals and operational reliability are not assessed. No probability of correctness is calculated.