Audit implementation and verification

The supplied audit targets 61322f73a77b7653ea8f2a8d3e10f0326a524414. Corrections were delivered as separate chapter/problem commits to main. The content record summarizes the instructional changes; this record also includes the reader and verification-tool corrections.

What to trust

A build assignment is not a finished service. An executable local reference is not a cloud deployment. The catalog distinguishes 18 chapters, 42 coding bundles, 49 project entries and 41 design/architecture pages. The project entries are 40 build assignments, five continuing-project stages and four runnable AI references. Tests do not demonstrate a learner's independent mastery.

Audit area	Implemented correction / retained boundary	Current check
PS-01–04, DS-01–04	Contract-based review; valid unchanged approval; binary-search boundaries; numeric overflow and supported inputs	Coding references and their regression tests
BE-01–04, DB-01–02, FE-01–03	Atomic decisions, cancellation limits, database-specific semantics, draft ownership and input validation	Python API tests; separate TypeScript and real HTTP/browser checks
TE-01–04	Registered suites; structured execution/skip results; failing omission/empty-suite controls; historical membership separate from current correctness	Runner regressions, catalog and preservation reports
SE-01–04, DA-01–03	Server-side authority, bounded authorization freshness, sandbox separation, ordering and redelivery distinctions	Focused references plus explicit design failure traces; no live security certification
SY-01–06, DE-01–02	Honest artifact scope, atomic assignment/booking/replay, cache expiry, admission versus recovery and stable rollout salt	Corrected design contracts; executable rollout example; diagram review
IN-01–02, OB-01–04, RE-01–03, PE-01–03	Configuration contracts, supported metrics components, causal links, dimensional bounds, retry budgets, fenced result pointers, scheduler catch-up and measured performance	Lessons and failure traces, reference suites, checked arithmetic
AI-01–05	Whole-field amounts, immutable published bytes, review versus retry, distinct rollback history, evaluated-candidate identity	AI reference and mocked-adapter tests; mocks are not AWS
MI-01–02, TD-01–04	Safe hard-case rehearsals, justified live pilots, staged benefits, valid approval without edits, evidence-based strategy and role expectations	Corrected lessons and linked project rubrics
SITE-01–08	Atomic publication, retained old assets, validated storage, locked dependencies, render-aware cache, exact content checks and usable contextual links	Publication failure tests; reader build, links, browser and completeness checks
META-01–02, VIS-01	Exact current catalogs, commit-scoped evidence, explicit inspected-versus-pending visual status	Catalog, Python, reader and visual manifests; no universal green certification
EVID-01–02	Original practice versus interview sightings; unknown dates remain unknown; public assessor keys are procedural separation	Existing source/exposure disclosures retained; no fresh online verification claimed
IN-03	Local/mocked status remains separate from live cloud verification	AWS workbench; explicit account authorization required

Reproduce the repository checks

From the root, with the documented dependencies installed:

python scripts/check_catalog.py --report /tmp/catalog.json
python scripts/check_organization.py --report /tmp/preservation.json
python scripts/check_visuals.py --report /tmp/visuals.json
python docs/check-links.py
python scripts/check_curriculum.py --report /tmp/python-results.json

The catalog report identifies each current source path and its hash. The Python report records source commit, committed tree, dirty-worktree status, runtime, commands, executed methods, skips and failures. A dirty-worktree result must not be represented as a test of the clean recorded commit. Missing/empty required suites fail rather than disappearing from the total.

The preservation report checks that the 472 distinct destinations of 477 historical source mappings remain present; several navigation sources were intentionally merged. Additions are allowed. Changed bytes are listed, not certified correct. check_organization.py --historical retains the original reorganization-only equality experiment; run it at the original reorganization checkout with the baseline Git objects available, not as a current-content gate.

Recorded reader run

At 63f663f49c87257c1c887167ba7bceb875dddde7, Actions run 35931170248 completed the reader build and tests before publication:

16 site regression methods passed, including publish failures and corrupt storage.
403 Mermaid diagrams rendered with zero failures; 320 content pages built.
108,958 local references across 322 HTML outputs resolved.
131 linked code files and all 292 source SVGs were checked in the output.
19 Chromium scenarios passed, including keyboard/touch prerequisite round trips.

These are the numbers for that exact run, not thresholds for future builds. Later code, content or renderer changes require the corresponding gate again. The generated content-inventory.json maps source pages to hashes, required diagrams, linked code and assets; it rejects missing required content even when an unrelated total remains large enough. Reference fixtures also verify that an empty output and missing headings/assets fail.

Visual evidence

Fifteen comparison diagrams had truncated text and vertical arrows suggesting that independent cases were sequential. Their repaired renders were inspected as three contact sheets; each row now reads failure → repair with complete labels. The retry diagram was inspected separately at full size: three total attempts per layer permit up to 27 leaf calls, not three retries and not a claim of 27 simultaneous calls. The exact inspected bytes are in visual-review.json (download file, source below).

Read the supplied code · visual-review.json

visual-review.json · visual-review.json

{
  "assets/design-interview/ad-click-aggregator-before.svg": {
    "sha256": "d455cd1505117f514c6429333f6c9038dd1ed0d5295f8833cb250d2068c8a6f8",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/api-gateway-platform-before.svg": {
    "sha256": "b1358eb9179bdc4ae362fc918fc5b730bc143310d97e17a06e37df5dc0d2867f",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/calendar-availability-before.svg": {
    "sha256": "68394ea7ca4e52432636cb0e38ed1bb5bef61021bf28fd946b3d6c84ab9da97c",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/distributed-cache-before.svg": {
    "sha256": "9931706b54d962d7b610bf9df9b489891b6fdf230a572b4e0efa7c75ea634837",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/distributed-key-value-store-before.svg": {
    "sha256": "91ff80948ebc073b9ec7de0800f47c5dcc6b39eb739410e73a93f66e7b163859",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/food-delivery-marketplace-before.svg": {
    "sha256": "fa3ba00e03763d4f0408b360f1d7d93ded3841e609ac42571a5ce58fb0a87931",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/job-scheduler-before.svg": {
    "sha256": "5499aac03facc44c0c3be39558c1ee33b034487b55794bd130d8c58e5bbf4631",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/metrics-platform-before.svg": {
    "sha256": "10ed2ff527dd4ce956526dd8e33e458f4d56ffde027a2e7f927b5b203b1d5dbb",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/news-aggregator-before.svg": {
    "sha256": "9ff110f8d7f6a124da6a07295721eb2f9e03c90660ca054915c28b3ccc1d68d1",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/online-judge-before.svg": {
    "sha256": "e324ace25a6ae373709191b7be77d512ebfc86293619c8107b9db825d6ca0adf",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/rideshare-dispatch-before.svg": {
    "sha256": "59d36931ec391cbcb9450b21e06566b6cf4628e201089e87bb01acdcfd703ded",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/typeahead-search-before.svg": {
    "sha256": "d002823c549407904b8790c2bc65532f7ac2df83b3c1e970861df87d5111c767",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/url-shortener-before.svg": {
    "sha256": "18265ace4c3f15f185b7a8e91c00ec9decccc934bb65282216ed532dc5a55265",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/video-streaming-platform-before.svg": {
    "sha256": "d48629413128fcd6f58f0e9aad7046110210127c76a6f62b4d5c15e6c0ab004e",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/design-interview/web-crawler-before.svg": {
    "sha256": "3b44caf91cb81d482eed2cca4e66f4663d4095e7dc735167f59c45466b524c1f",
    "review": "Static render and three independent failure/repair rows inspected; not a deployed-system test."
  },
  "assets/diagrams/retry-amplification.svg": {
    "sha256": "74932c8961d8266b68d076d43bc31e48bcad8b06c0907bb19cd007eb1d0b4208",
    "review": "Static render inspected: three total attempts per layer permit up to 27 leaf calls; no claim of simultaneous retries."
  }
}

check_visuals.py gives every tracked SVG and Mermaid block an explicit status. A changed inspected SVG invalidates that inspection. Unlisted diagrams remain semantic-review-pending; successful parsing/rendering and unchanged bytes do not silently turn them into reviewed diagrams. Animated visuals retain a separate playback/static-equivalence gate. These labels describe evidence, not a claim that pending visuals are defective.

Separate gates and limits

The verification workflow reports Python references, reader/browser/TypeScript, and real PostgreSQL experiments separately, with logs tied to its commit SHA. Read the individual job results; a completed job name alone is not evidence that a different gate ran. The reader artifact includes the exact content/visual manifests, rendered browser screenshots and command output.

No live AWS resources or production model evaluations were run for this audit. Cloud IAM, regional availability, quota/cost behavior, notification destinations and versioned-resource cleanup still require the authorized disposable-account exercise. Do not infer them from mocks or static templates. No new hiring probability, interview-frequency ranking, all-browser guarantee or independent learner score is asserted. This is an implementation record, not blanket certification that every one of the audit's optional external gates has run.