200-901 CCNA Automation deep review — September 29, 2026¶
The guide now maps all 61 numbered objectives and their supporting bullets from the actual three-page v1.1 blueprint. It corrects a Boolean-parsing defect and a single-page inventory example, adds an executed Requests fixture workbook with 67 checks, and provides 48 original answered readiness prompts. Eight infrastructure labs and independent human review remain pending; this is a same-context AI review using public sources.
Scope, lifecycle and source conflicts¶
The exam page still labels the exam v1.0, while the three-page blueprint and four-page training PDF say v1.1. Both actual PDFs were fully read, with binary hashes and page receipts recorded. The six domain counts are 8/9/9/12/14/9; weights remain 15/20/15/15/20/15. Supporting bullets are included in the mapping, not counted as additional numbered objectives.
Current canonical logistics are 120 minutes, USD300, English/Japanese. The credential page corroborates duration/price, no formal prerequisites and three-year validity but lists English only. The conflict remains visible; no scheduling checkout was checked. Current Associate policy permits the 30-CE route before expiry. Training recommendations and its advertised 48 CE are not exam prerequisites or proof of an individual's earned credits.
The May 27, 2025 Par Merat announcement and August 6, 2025 Quinn Snyder explanation were read for their Associate transition text. The latter explicitly separates the Associate/Expert rename from the Professional redesign. February 3, 2026 branding is retained without importing CCNP objectives. The current detailed PDF controls scope; broad portfolio wording is insufficient evidence of additional Associate requirements.
Reviewed monitor difference and historical evidence¶
The initial objective digest changed from b741522983500658e43c1a9a0200c142e4a626ff8d0ad8a01859f92d0670c60f to 389efdb0eae2c7136ded47c44e8dd55bf4076e03a130ff75ce3d49b49224545d. An exact comparison found 35 leading title/navigation lines before an otherwise identical previous exam-content block. This establishes an extraction difference, not a new exam scope or when Cisco changed its page. The generic extractor was not changed.
The earlier snapshot was preserved byte-for-byte before explicit acceptance. The previously absent lifecycle baseline was initialized at c9a9888033258e9e81e3c9c7127203f73394de2322f1b230047288f0215528a9. The operational record retains initial, accepted and subsequent observations. The original September 2 review and historical audit retain their dates, hashes and conclusions with the archived snapshot path; the current review has a new dated ID.
Applied teaching corrections¶
The old parser returned reachable=True for the string "false" and raised AttributeError on an array root. Both were reproduced using its exact original code and synthetic inputs. The replacement validates object shape, required fields, nonblank string ID and exact Boolean type; it deliberately rejects duplicate JSON keys and non-finite constants. Extra-field rejection belongs to the fictional contract rather than every production API.
The former Requests example returned only the first page. The new workbook exercises prepared GET requests, next links, schema validation, repeated identical records, conflicting records, cycles, page/record/request limits and a later-page denial. It returns the collection only when all pages succeed. It also distinguishes JSON decoding from HTTP success, redirects, content type, 204, auth/permission failures, conditional conflicts, rate limits, selected transient failures and terminal TLS errors. Retry-After may be seconds or an HTTP date; a delay outside the remaining wait budget causes failure rather than premature retry.
Provider-specific explanations use Meraki pagination, rate-limit, error and SDK documentation. The invented inventory.invalid array schema is not presented as a Meraki/Cisco endpoint. SDK automation does not establish complete inventory or safe policy, and no SDK was installed.
Selected NETCONF RFC 6241 sections distinguish optional candidate/confirmed-commit/rollback capabilities, shared candidate locking, persistence and application health. Selected RESTCONF RFC 8040 sections explain resource/datastore ETags and conditional edits. Ansible check mode can skip unsupported tasks or execute tasks marked check_mode: false; sensitive diffs need protection. Docker ARG/ENV are unsuitable secret stores. None of those infrastructure operations was executed.
Webhook prose now requires the provider-defined signature over original bytes and signed freshness only when the contract supplies it; signatures alone do not prevent replay. The current Webex page returned a minimal shell and browser extraction failed on size. Search snippets were insufficient to assert complete current signature/delivery behavior. A separate 2017 Cisco blog was inspected but not adopted as current contract evidence. No Webex receiver, secret or message was created.
Executed evidence and its limits¶
The exact public workbook was executed with Python 3.13.14 and already-installed Requests 2.34.2: 67 local checks passed. Its code digest, stdout, interpreter version, original-defect receipt and detailed boundary are recorded in the operational evidence file ADLC_Docs/operations/2026-09-29-200-901-deep-review.json in the repository.
A Requests BaseAdapter replaces transport for both HTTP schemes. Session preparation, response objects and link parsing run; no DNS, socket, authenticated API, actual sleeping or real credential is used. Fixed fixtures supply a literal inert authorization header and virtual waits. The two-page case produces devices a and b; the final printed retry case produces device a, Boolean false, two requests and virtual wait [2].
The body cap is checked after buffering, not during a streaming download. The wait budget is not an overall deadline, and connect/read timeouts are not a whole-operation deadline. Link validation covers selected relations and malformed fixtures, not all RFC ambiguities. There is no cross-page snapshot guarantee, write-idempotency solution, durable storage, webhook implementation, general SSRF proof or production-readiness claim. Private response-buffer attributes are used only for fixture construction. Python JSON documentation was currently 3.14.7; the executed interpreter was 3.13.14.
Eight proposed labs specify inputs, failure injections, expected evidence, permission boundaries and cleanup/restoration. The three retained scenarios remain plans. The 48 answers are original teaching explanations, not obtained exam questions.
Learning-content findings¶
The April 13, 2026 Margaret Dilloway article was fully read and reports 32h30 for the Cisco learning path. This is a dated public duration, not a fresh authenticated path sum. Public training HTML/PDF contain 13 outcomes, 14 outline parts and 18 labs; the Singleton training lab does not replace the exam's MVC/Observer objectives.
LinkedIn's public metadata reports 13h02m, March 22, 2024 and Kevin Wallace/Charles Judd. Selected early outline sections were inspected; paid lessons, transcripts and exercise files were not. The former Pearson second-edition URL now redirects to an empty search response, so publication/availability are unverified and the earlier forthcoming September 15 statement was withdrawn. Search snippets and this repository's published guide were not independent proof of availability. O'Reilly/Udemy returned HTTP403; previous duration/update claims are explicitly unverified. YouTube returned only a title/footer shell.
The complete public Sandbox overview distinguishes shared nonadministrative Always-On environments from private reservations with VPN/admin/setup/expiry constraints. No account, reservation or live availability was checked. Catalog runtimes, study budgets, course access and quality are kept separate.
Validation and remaining work¶
Repository tests, repository validation, learning-catalog consistency, strict site build, generated-site validation and diff checks are recorded in the operational evidence when complete. Historical evidence is preserved rather than relabeled as newly executed work. Remaining blockers are live infrastructure exercises, selected inaccessible or conflicting public information, and independent human review. No runtime installation, account authentication, service change, notification or edit to the five reserved GitHub guides was made.