~ peterdongo_

// anthropic exams

Claude certification exams

Anthropic runs a small family of Claude certifications - an Associate track for business users, a Developer track for people building on the API, and a two-level Architect track for designing systems around the model. While preparing for them I wrote four full-length practice exams with Claude, on my own account, and I'm hosting them here along with the tips and materials that actually helped. They're original, unofficial, and not affiliated with Anthropic - but they mirror the real format closely.

// format

How the exams work

  • Multiple choice and multiple response. Every item tells you how many answers to select. Multiple-response items only score when you pick exactly the right set.
  • Scaled scoring. Your raw score maps to an approximate 100-1000 scale, with the pass cut at 720, plus a percent-correct breakdown per domain - just like the real score report.
  • Timed, 120 minutes. Generous for the item counts, but the scenario-based Architect questions eat clock. Read carefully.
  • Blueprint-weighted. Questions are distributed across the published domain weights, so a weak domain shows up clearly in your results.

// practice

The practice exams

CCAO-F Foundations

Claude Certified Associate

For business users of Claude. No API, agent, or coding knowledge needed - prompting, evaluation, governance and workflow design.

60 items · 120 min · pass 720 Start →
CCDV-F Foundations

Claude Certified Developer

The developer track: the Messages API, tool use, prompt engineering in code, streaming and building on Claude.

53 items · 120 min · pass 720 Start →
CCAR-F Foundations

Claude Certified Architect

Scenario-based. Designing reliable systems around Claude - agents, context, validation and failure handling.

60 items · 120 min · pass 720 Start →
CCAR-P Professional

Claude Certified Architect

The hardest of the four. Deep, multi-part scenarios on production architecture, trade-offs and enforcement.

63 items · 120 min · pass 720 Start →

Each exam runs entirely in your browser - nothing is submitted anywhere. Your answers, timer and score stay on your machine.

// tips

Tips that move the needle

The meta-lesson underneath all of these: you're building a reliable system around a fallible model. Almost every right answer follows from that.

  • 01

    Constrain, don’t add.

    “More tools / more agents / a bigger context window” is almost always the wrong answer. The right fix usually removes a degree of freedom rather than adding one.

  • 02

    Prompts suggest; systems enforce.

    If a rule must always hold - money, compliance, ordering - enforce it in code or a hook, not by asking the model nicely.

  • 03

    Fix the earliest layer.

    Most wrong answers are reasonable fixes applied too late in the pipeline. Trace the failure back to where it first became possible and fix it there.

  • 04

    Match the fix to the failure.

    Vague rule → explicit criteria. Inconsistent output → few-shot examples. Invalid JSON → a schema. Valid-but-wrong output → semantic validation.

  • 05

    Don’t collapse states; escalate on policy, not vibes.

    “Tool failed” is not “found nothing.” Escalate on a policy or workflow trigger - never on sentiment or the model’s stated confidence.

  • 06

    Remember what you’re building.

    You’re building a reliable system around a fallible model. So “just fine-tune it” or “tell it to be more careful” are almost never the answer.

// prep

How I prepared

What I actually leaned on, beyond these practice exams. The other mock exams below are external - I found them genuinely useful for getting a feel for the question style.

➜ These practice exams are unofficial and not affiliated with, endorsed by, or produced by Anthropic. They were generated with Claude as study aids and are written against the published exam guides - not drawn from any live question bank. “Claude” and “Anthropic” are trademarks of Anthropic.