06-reference/research

anthropic cert public coverage gap

2026-07-17·research-brief·source: deep-research·by Ray Data Co (deep-research synthesis)
certificationanthropiccca-foundationsphdatastudy-plan

The Portal Never Gated the Curriculum — It Gated the Exam Seat, and That Gate Is Already Open

The question

What % of the Anthropic Partner Network certification curriculum is covered in publicly-accessible Anthropic surfaces (docs, cookbook, courses.anthropic.com, claude.ai blog, YouTube), and what's the actual gap that requires portal access — measured by sampling 5-10 known Partner-cert exam topics against public material?

Auto-promoted 2026-05-22 (score 12/15) to test the founder's gut estimate that "~70% of the cert content is public," in service of a portal-blocked study path for the +$5k CCA-Foundations escalator (target 2026-11-22).

What we already know (from the vault)

What the web says

Convergences and contradictions

Synthesis for RDCO

The percentage is not measurable as asked, and the framing is wrong in a way that matters more than the number. The question assumes the cert curriculum is a body of knowledge scattered across docs/cookbook/blog/YouTube, and that the portal holds some withheld remainder. That is not the shape of this cert. The curriculum specification — the actual denominator — lives in a single artifact, the official exam guide, which was never distributed through docs.anthropic.com or YouTube at all. It is linked from the cert landing page and RDCO has held a copy since 2026-05-27 (v0.1 via EAP) and the current v0.2 since 2026-07-02. So the honest answer to "what % of the curriculum is public" is: the denominator was never enumerable from the surfaces the question names, and it never needed to be — we have it directly. I did not sample the 27 task statements one by one against public docs, and I am explicitly declining to produce a coverage percentage. Any number I generated here would be invented. The founder's "~70%" was a reasonable gut call answering a question that turned out not to be the operative one; it should be neither validated nor corrected, just retired.

What the portal actually gates, precisely: not the exam guide, and not the Anthropic Academy learning path (four courses, free Skilljar account, explicitly no Anthropic account required — so the "partner-gated courses" assumption is false). What it gates is the exam seat itself — the $125 purchase behind anthropic-partners.skilljar.com, requiring a partner company email, now routed through Pearson VUE. That is a ~15-minute account-creation action with a phData email, per a colleague who has already walked it end to end. It is a registration step, not a study blocker, and it was never on the critical path for learning.

Therefore the requested deliverable — "a working portal-blocked study path" — should not be built. It would be a strictly worse artifact than what the vault already contains. The real study path is [[2026-05-27-claude-certified-architect-foundations-study-plan]] (phase plan against official weights) plus [[2026-07-02-cca-exam-guide-update-analysis]] (27 task statements, 12 worked sample Qs, out-of-scope cuts) plus [[cert-progress]] (live weak-axis tracking). Building a public-surfaces-only substitute would reintroduce exactly the third-party inaccuracies flagged above — the same failure mode that produced the stale [[anthropic-certification-study-guide]], which was built from a third-party article and now carries a wrong domain taxonomy and an out-of-scope streaming drill.

Is the portal blocker fatal to 2026-11-22? No — and the deadline is not the live question anymore. At 83% cold on the first practice exam, the vault's own read is that the founder should target the 90-day escalator (~2026-08-26), not the 180-day one. The binding constraints are (1) the exam-2 dip to 70% borderline, traced to three cheap patterns — test-craft on "BEST" stems, hook scoping, and API surface — not broad weakness, and (2) actually creating the Skilljar account. The single highest-value action is not more research: it is the 15-minute account creation, which either clears the gate or finally reveals the real blocker. The one concrete research-side action this brief does justify is hygiene: mark [[anthropic-certification-study-guide]] status: superseded, since it is the only artifact in the study set that would actively teach wrong material — a recommendation the 07-02 analysis already made and which is still unexecuted 15 days later.

Why this is in the vault

This brief closes an open Research Backlog item (Notion 363f7d4936d181499c6cfd2d42bfc65d) by falsifying its own premise, and it records the audit trail for the CCA-Foundations escalator decision: the portal was never the study blocker, so the 2026-11-22 target should be re-cut toward the ~2026-08-26 90-day escalator. It also documents that public third-party CCA prep material is reliable on domain weights but wrong on exam mechanics — a reusable trust boundary for any future cert study path, and the specific reason [[anthropic-certification-study-guide]] must be retired.

Open follow-ups

Related

Sources

Vault

Web