Practice for Claude Certified Developer – Foundations against its published blueprint, end to end: scenario items at exam register, flashcards scheduled by spaced repetition, and explanations written against the sub-skill you actually missed.
Exam figures published by Anthropic in the CCDV-F exam guide, v1.0, effective July 2026.
A support agent exposes four tools. Users who ask to open a ticket keep receiving documentation instead. The tool descriptions each read as one line and share vocabulary. Which change most reliably corrects selection?
// agent.tools[] — as deployed
{ "name": "search_docs", "description": "Find help content for the user" }
{ "name": "create_ticket", "description": "Help the user with their issue" }
{ "name": "lookup_order", "description": "Get information for the user" } Forcing tool_choice repairs one request and breaks every other path through the workflow. Descriptions that name their trigger and their boundary fix selection wherever the tool appears.
Synthay item, written to the published blueprint.
Not an Anthropic exam question. Session state and timings illustrative.
Active recall is the easy half. What separates a question bank from a preparation engine is what happens in the two hundred milliseconds after you get something wrong — and in the six days after that.
“Tools and MCPs” is 10.6% of the published exam and three separate competencies. Synthay records the miss where the exam scores it.
A rationale that only restates the right answer teaches nothing. The explanation starts from the reasoning that pulled you to A.
tool_choice is a per-request instrument. Pinning it to create_ticket repairs the ticket path and breaks lookup and escalation on the same agent. Selection failures that come from overlapping descriptions are fixed in the descriptions: name the trigger condition, and name what the tool is not for. The fix then holds wherever that tool is mounted.
A missed card drops to the front of the ladder and climbs again only when you answer it cold. Cards you already own get out of your way.
The exam guide states your score report gives percent-correct by domain. Your readiness view is built the same way, so a weak domain is visible before exam day, not after.
Illustrative session data. Domain weights and score-report format from the CCDV-F exam guide, v1.0.
The CCDV-F guide states the approximate share of scored items in every domain, resolved to twenty-five sub-skills. Synthay’s paths follow those numbers rather than a syllabus of our own invention.
Tile width follows exam weight; the two lightest domains are widened to stay legible. Every sub-skill listed has Synthay questions.
Domain and sub-skill weights as published in the Claude Certified Developer – Foundations exam guide, v1.0, effective July 2026. The guide notes that domain percentages are not themselves used to determine a pass or fail result.
The program publishes four exam guides. Synthay is built for Claude Certified Developer – Foundations and works its blueprint in full. The rest are listed as published, and no prerequisite links them — each is sat on its own.
For engineers who build, integrate and ship Claude applications, agents and workflows. Eight domains, from Applications and Integration at a third of the paper down to Debugging at 2.6%.
Exam figures are Anthropic’s, from the CCDV-F guide v1.0. The fee is paid to Pearson VUE, not to Synthay.
For non-developer professionals: prompting, output validation, product selection, governance. No API or software-development experience assumed.
60 items · 120 min · $99
Agentic orchestration, tool and MCP interfaces, Claude Code configuration, context reliability. Five domains, twenty-seven task statements.
60 items · 4 scenarios from a bank of 6 · $125
End-to-end solution design, retrieval pipelines, evaluation and observability, governance and stakeholder lifecycle. Seven domains.
63 items · 120 min · $175
Specifications as published in each credential’s exam guide, v1.0, effective July 2026. All four are proctored by Pearson VUE, run 120 minutes, pass at a scaled 720 of 100–1,000, and stay valid 12 months. None lists a mandatory prerequisite.
Preparation that fits between a standup and a deploy: driven from the keyboard, resumable mid-session, and honest about how much of the paper you have left.
Every action in a session has a key. Answer, reveal, flag, skip, move between items and close out — without reaching for a pointer once.
Fifteen items, thirty minutes, no explanations until you submit. Items are drawn per domain in the published proportions, and come in both formats the guide lists — multiple-choice and multiple-response, which states how many responses to select — so a short form carries the shape of the real paper without asking two hours of you.
Item counts derived by applying the guide’s published domain weights to a 15-item form. The two lightest domains, at 3.1% and 2.6%, round to zero at this length.
Weights, item counts and policies on this page are quoted from the published exam guides, with the version they were read from. When a guide revises, the questions and this page move with it.
| Credential | Code | Guide | Effective | Coverage |
|---|---|---|---|---|
| Associate – Foundations | CCAO-F | v1.0 | Jul 2026 | — |
| Developer – Foundations | CCDV-F | v1.0 | Jul 2026 | full |
| Architect – Foundations | CCAR-F | v1.0 | Jul 2026 | — |
| Architect – Professional | CCAR-P | v1.0 | Jul 2026 | — |
Coverage “full” means every one of the guide’s twenty-five sub-skills has questions written against it, sampled per domain at the published weights — 180 items today. It is not a claim to reproduce the live item bank.
Exam policy below is quoted from the published guides. Where the answer is about Synthay rather than the exam, it says so.
Something not covered? Ask usNo. Synthay is an independent preparation product, not affiliated with, endorsed by, or authorized by Anthropic. Everything quoted here comes from the publicly available exam guides, cited with the version it was read from.
No, and they never will be. Every item in Synthay is written by us against the published blueprint. Candidates accept a confidentiality agreement covering all exam content before the exam begins; live items are Anthropic’s confidential property, and reproducing them would put your credential at risk as well as ours.
No. None of the four guides lists a mandatory prerequisite or required course, and no ordering between the credentials is published — Architect – Professional does not require Architect – Foundations. The credential is awarded on exam performance alone.
For CCDV-F the guide recommends, without requiring: one to five years in software engineering, six months or more hands-on with Claude or comparable LLM systems, proficiency in Python and/or TypeScript, and fluency with REST APIs and CLI tools.
Results are reported on a 100–1,000 scale with a cut score of 720. The exam is criterion-referenced: you are measured against a standard set by a formal standard-setting study, not against other candidates. The published domain percentages describe how items are distributed — the guide states they are not used to determine your pass or fail result.
Your score report gives pass or fail, the scaled score, and percent-correct by domain.
You may retake after a waiting period that lengthens with each attempt: 14 days after the first failure, 30 after the second, 90 after the third. You may sit a given exam up to four times in a rolling twelve months, and the fee applies to each attempt. Limits are per exam, so not passing one does not block you from registering for a different one.
Twelve months from the date it is awarded. Renewal is a free, non-proctored assessment on the Anthropic Partner Academy, with no fee for renewing on time. If the credential lapses you must retake the full exam at the full fee to regain certified status.
Pearson VUE, online proctored and/or at a test center per program policy, with registration through the Anthropic Partner Academy. CCDV-F is $125 USD; Associate – Foundations is $99, Architect – Foundations $125, Architect – Professional $175. You can cancel or reschedule up to 24 hours before your appointment; changes inside 24 hours forfeit the fee.
We won’t promise that, and neither will anyone honest. The guide is explicit that Anthropic does not guarantee any particular resource ensures a passing result, and it recommends combining study with building and operating a real Claude application. Synthay covers the blueprint and shows you where you are weak; the hands-on part is still yours.
The first session takes about twelve minutes and ends like this — a readiness figure for all eight domains, with the weakest one a click from its own drill. No card details, no trial countdown.
Independent preparation. Not affiliated with Anthropic.
Readiness by domain
Your weakest sub-skill — D8 · Tool Implementation — is one click from its own drill.
Illustrative session data.