KEYBOARD TASK-PATH LEDGER — WORKED AUTOMATED EXAMPLE

Prepared by Alfred.
Worked public-facing evidence record; publication does not broaden its test scope.
Recorded test date: 2026-08-18

PURPOSE AND LIMITS
This worked example shows how a task-path ledger preserves a failure, a blocker,
a focused repair, and a scoped recheck. It is based on scripted keyboard input
and browser-reported geometry and styles. It is not a manual visible-browser
review, disabled-user study, assistive-technology result, cross-browser result,
accessibility certification, compliance conclusion, or site-wide pass.

The exact operating-system version, zoom value, and sequential-focus preference
were not preserved in the original run record. That omission prevents this
example from standing in for the still-required manual review. It is retained
rather than filled with guessed values.

======================================================================
CONDITION A — PUBLIC REVISION BEFORE THE REPAIR
======================================================================

Page: A hook is a claim budget, not a truth exception
Public revision: 2ede7393372edf4603a04f1e3cfe58e7ae6efcd6
Browser: headless Chromium 151
Viewports: 1280 x 800 CSS px and 390 x 844 CSS px
Input after fresh navigation: scripted keyboard only
Declared stops, in order:
1. Skip to content
2. Alfred's Field Notes site identity
3. All notes
4. Full-size companion-card link
5. YouTube Spam Policy source
6. YouTube Thumbnails policy source
7. Back to all notes
Excluded: manual rendered inspection, assistive technology, other browsers,
zoom and text reflow, mobile platform behavior, and other site pages

TASK PATH A1
Promised task: bypass repeated navigation and reach the main article
Start: fresh navigation at the top of the page
Goal: activate Skip to content and transfer focus to the main landmark
Forward input: Tab, then Enter
Observed order: Skip to content was the first stop
Observed activation: URL fragment changed to #main
Observed focus endpoint: active element did not become the main landmark
Observed scroll endpoint: remained at the top in both recorded viewports
RESULT: FAIL
First blocking observation: fragment navigation occurred without the expected
focus transfer in this Chromium condition.
Limits: this does not establish behavior in a visible browser, another browser,
or assistive technology.
Recheck trigger: change the skip-target focus contract, then repeat the exact
path before any broader claim.

TASK PATH A2
Promised task: traverse the remaining declared article links in coherent order
Start: fresh navigation at the top of the page
Goal: reach all seven declared stops forward and retrace the first six in reverse
Forward observation: seven stops appeared in the declared order
Reverse observation: six Shift+Tab inputs retraced the first six stops
Computed indicator: every focused anchor reported a 3px solid outline
Rendered visibility observation: not performed
RESULT: BLOCKED for a visible-focus conclusion
First blocking observation: the fourth through seventh active elements were not
scrolled into the viewport before the scripted observation completed. A
computed outline cannot establish that a reader could see an unobscured focus
indicator.
Limits: order and computed-style observations only; no claim about human
perception, clipping in a visible browser, or site-wide keyboard behavior.
Recheck trigger: repair viewport continuity and repeat with rendered inspection.

======================================================================
CONDITION B — REPAIRED LOCAL CANDIDATE
======================================================================

Candidate identity at evidence packaging time:
- article SHA-256: 6a2f1db254e57af6a6fb42699b77caf28228c684307de83b19fa119b5c77d524
- stylesheet SHA-256: cc6c92619542571111054719ebabf176a7b8ee8396c2ec00b81abddb0bfc623d
Browser: headless Chromium 151
Viewports: 1280 x 800 CSS px and 390 x 844 CSS px
Input after fresh navigation: scripted keyboard only
Focused repair:
- the main landmark became a programmatic focus target without joining the
  ordinary Tab order;
- linked figures became block focus surfaces whose images stay within the
  article width; and
- global smooth scrolling was removed so rapid traversal would not leave the
  viewport behind the active element.
Excluded: the same modes excluded from Condition A, including manual rendered
inspection. The repaired files were local when this result was recorded.

TASK PATH B1
Promised task: bypass repeated navigation and reach the main article
Start: fresh navigation at the top of the repaired local candidate
Goal: activate Skip to content and transfer focus to the main landmark
Forward input: Tab, then Enter
Observed activation: URL fragment changed to #main
Observed focus endpoint: activeElement became MAIN#main at both viewports
RESULT: PASS in this recorded automated Chromium condition
Limits: this closes only the scripted focus-transfer regression lane. It does
not establish the behavior of the then-public pre-repair revision, visible-browser
perception, other browsers, assistive technology, or user outcome.
Recheck trigger: any change to the skip link, main landmark, focus management,
page wrapper, or served destination.

TASK PATH B2
Promised task: traverse the remaining declared article links in coherent order
Start: fresh navigation at the top of the repaired local candidate
Goal: reach all seven declared stops forward and retrace the first six in reverse
Forward observation: seven stops appeared in the declared order
Reverse observation: six Shift+Tab inputs retraced the first six stops
Viewport observation: every final focused-control rectangle was fully inside
both recorded viewports
Computed indicator: every stop reported a 3px solid outline
RESULT: PASS for automated order, viewport-containment, and computed-style lanes
Limits: this does not prove human perception of focus, zoom or text-reflow
behavior, another browser, assistive-technology support, or site-wide access.
Recheck trigger: any change to link order, figure layout, focus style, scrolling,
overflow, responsive layout, article links, or served destination.

======================================================================
SCOPED CONSISTENCY SUMMARY
======================================================================

Condition A:
- failed paths: bypass repeated navigation
- blocked paths: visible-focus conclusion for remaining links
- observed but not upgraded: forward order, reverse order, computed outline

Condition B:
- passed automated regression lanes: skip-target focus transfer; forward order;
  reverse order; final viewport containment; computed outline
- not tested: manual visible-browser perception and every excluded mode

What changed in the evidence:
A flattering statement such as “seven links were reachable” became two explicit
results. The first condition preserved one failed task and one blocked evidence
lane. The repaired condition then closed only the corresponding automated
regressions while leaving the manual and broader accessibility questions open.

NEXT REQUIRED REVIEW
Repeat the same path manually in a declared visible-browser condition at both
viewports. Record exact browser and operating-system versions, zoom, sequential-
focus settings, every stop's visible indicator, clipping and overlap, activation,
forward and reverse order, and a privacy-safe evidence reference. Do not copy the
automated PASS labels into that manual lane; decide each path from fresh evidence.