06-reference/research

ccdvf recertification exposure 2026 12 07 content change

2026-11-07·research-brief·source: deep-research·by Ray Data Co (deep-research synthesis)
anthropic-certificationccdv-frecertificationcert-policyphdata-cert

The clause exists in the CCDV-F guide word-for-word, the Terms go further than the guide, and Anthropic has not published the rule that would decide it

The question

"Does the 2026-12-07 Anthropic exam content change put the founder's already-passed Claude Developer Foundations credential under the same guide section 14 full-retake-to-recertify risk, since CCDV-F is on the same change list?"

Context: derivative of [[2026-11-02-claude-architect-pro-2026-12-07-content-delta]], which found §14 of the CCAR-P guide but did not check whether the same clause governs an already-earned credential. The founder passed CCDV-F on 2026-10-30 at 926/1000, on the outgoing content, 38 days before the change date. date checked 2026-11-07 01:12 EST.

What we already know (from the vault)

What the web says

All primary, all from Anthropic's own Partner Academy, all publicly readable without login, retrieved 2026-11-07 via Playwright and curl.

Convergences and contradictions

Synthesis for RDCO

Direct answer: yes on paper, no in practice, and the thing that would settle it has not been published. The CCDV-F guide contains §14 word-for-word identical to CCAR-P's, so there is no asymmetry between the two certs — the founder's already-earned credential sits under exactly the clause the parent brief flagged. The exposure is in fact slightly worse than the parent brief knew, because Terms §2 lets Anthropic require recertification before the 12-month term expires on "material updates," which the parent brief never surfaced. But all three clauses are permissive, none has ever been exercised, 12-07 is nowhere characterised by Anthropic as substantial or significant, and the public FAQ concedes outright that "Full details will be shared before the first certifications come up for renewal." The honest finding is the negative one: Anthropic does not publish the answer to this question, and the specific page I checked says the rule is still being written. Anyone converting these clauses into a confident forecast — in either direction — is inventing policy.

The backlog entry's framing should be walked back one notch. It said "a banked escalator leg is not as banked as cert-progress.md records it." Two corrections. First, the $500 bonus and the deal placement attach to having passed — that money is paid and the pass is a historical fact that no recertification rule reaches. What is exposed is whether the credential stays current past ~2027-10-30, which is a different and much cheaper asset. Second, cert-progress.md does not actually record an expiry at all, so there is no stale claim to correct; there is a missing field. The concrete edit is at /Users/ray/rdco-vault/01-projects/phdata/cert-progress.md line 238 (the 2026-10-30 pass record), which should carry the expiry and a one-line renewal note, and the surrounding lines 221-222 (the 2026-10-21 sit-before-12-07 decision), which should carry a pointer to this brief so the reasoning behind that date is recoverable. Worst-case financial exposure if the clause bites at renewal is one $125 proctored sitting in roughly October 2027, against a free open-book assessment otherwise. Whether phData reimburses a renewal rather than a first attempt is (U) — the reimbursement policy noted in [[cert-progress]] 2026-10-21 was never verified and is scoped to attempts, not renewals.

The sequencing implication is narrow and it is the opposite of alarming. The renewal mechanism as published is "review what's changed since you certified and complete a free, non-proctored assessment." That assessment is already scoped to the delta since his cert date — which by construction includes the 12-07 content. If Anthropic honours that mechanism, the 12-07 change is absorbed for free and §14 never fires. §14 only matters if Anthropic decides the delta is too large for an open-book assessment to measure, and the directional precedent runs the other way: the last time the program changed structurally, Anthropic extended 6-month credentials to 12 months automatically. So nothing in the founder's 2026 cert plan should move on this. CCAR-P on 12-22 stays right, for the reasons in the parent brief minus the §14 argument, which is weaker than that brief claimed. Do not re-litigate the 11-10 CCDV-F sitting — he passed at 926, the credential and the bonus are banked, and a pre-change pass does not forfeit anything that a post-change pass would have kept.

The one real action is a dated observation, not a decision. On 2026-11-17 the updated CCDV-F guide publishes. Its §16 Document Control table will say whether Anthropic calls this a new version (which would connect it to the policies page's "when the exam moves to a new version" language and raise the odds of the substantial-update reading) or a revision to v1.0. That single table row is the highest-information, lowest-cost signal available on this question, and it arrives in 10 days. Everything else here is reserved-rights boilerplate that three separate Anthropic documents decline to make concrete. The pre-change baseline is now archived locally (guide, Terms, and Exam Policy, all three with SHA-1s in Sources) so the 11-17 diff is actually performable rather than hypothetical — the parent brief recommended this for CCAR-P and it is now done for CCDV-F as well.

Why this is in the vault

It closes the open follow-up that [[2026-11-02-claude-architect-pro-2026-12-07-content-delta]] left unanswered ("Does the 12-07 change put his CCDV-F credential under the same §14 risk?") with primary-source quotes rather than inference, and it corrects that brief's over-reading of §14 as a penalty on sitting date. It also supplies the exact edit target — [[cert-progress]] line 238 plus lines 221-222 — for recording the founder's CCDV-F expiry and renewal exposure in the tracker that feeds the /cert-study weekly pulse and goals KR 4.1, and archives the pre-change CCDV-F guide, Terms, and Exam Policy so the 2026-11-17 refresh is diffable instead of silently overwriting the only baseline.

Open follow-ups

Related

Sources

Primary (Anthropic, retrieved 2026-11-07, publicly readable without login)

Vault

Third-party (consulted via one WebSearch, NOT relied on for any claim above; listed for traceability)