Four exams,
27 weighted domains.
Every domain below is transcribed from the official exam guide, weight included. Certly uses those weights twice: to build a simulation with the right shape, and to turn your answers into a projected score the same way the real report does.
Certified Associate
Foundations
Using Claude well as a knowledge worker: prompting, judging and validating output, picking the right product and model, and applying governance and responsible-use rules.
Audience: business and technical professionals using Claude day to day.
Certified Developer
Foundations
Hands-on building and shipping: API mechanics, agents and workflows, model selection and cost, prompt and context engineering, security, tools and MCP servers.
Audience: engineers building and integrating Claude applications.
The two smallest domains carry more questions than their weight implies, and that is deliberate. A three-question domain would be trivial to exhaust. Scoring still uses the official weight, so the extra practice never inflates your projected score.
Certified Architect
Foundations
Architectural judgment for production systems: agentic loops, multi-agent orchestration, MCP tool design, AI coding tool configuration, structured output and context management.
Audience: solution architects designing production applications. This is the exam Certly shipped with, and still the deepest bank.
This is the one exam built around scenarios. The real thing draws four from a bank of six, and every question sits inside one of them. Certly mirrors that: 114 of its questions are tagged to a scenario, and you can study any scenario end to end.
Customer Support Resolution Agent
Returns, billing disputes and account issues through custom MCP tools, plus knowing when to escalate.
Code Generation with AI Tools
Generation, refactoring, debugging and docs with custom slash commands, config files and plan mode.
Multi-Agent Research System
A coordinator delegating to search, document analysis, synthesis and reporting subagents.
Developer Productivity Tools
Exploring unfamiliar codebases, explaining legacy systems, automating repetitive work with MCP servers.
AI Tools for Continuous Integration
Automated review, test generation and pull request feedback where false positives are expensive.
Structured Data Extraction
Pulling fields from unstructured documents, validating against JSON schemas, handling edge cases.
Certified Architect
Professional
Owning a Claude system end to end: solution design, integration and RAG, evaluation and optimization, governance and risk, and communicating architectural trade-offs to stakeholders.
Audience: mid to senior-level architects owning systems in production.
A 90% in a 3% domain is worth almost nothing.
Raw accuracy flatters you. Certly multiplies your accuracy in each domain by that domain’s official weight, then maps the result onto the 100 to 1000 scale with 720 as the line. That is why the dashboard leads with a projected score instead of a percentage. It is the number that decides the outcome.
Domain names, weights, item counts, time limits and fees come from each exam guide, version 1.0, effective July 2026. Certly’s questions are written independently and are not real exam items.
Pick your exam and see where you actually stand.
