This certification validates that practitioners can make informed tradeoff decisions when shipping real-world solutions with Claude. It tests foundational knowledge across the Claude API, the Claude Agent SDK, Claude Code, and the Model Context Protocol (MCP) — the technologies used to build production-grade applications with Claude.
Questions are grounded in realistic scenarios drawn from actual customer use cases: agentic customer support, multi-agent research pipelines, Claude Code in CI/CD, developer productivity tools, and structured data extraction.
The ideal candidate designs and operates production applications with Claude and understands both the capabilities and the limitations of large language models.
Multi-agent orchestration, subagent delegation, tool integration, lifecycle hooks.
CLAUDE.md, Agent Skills, MCP servers, plan mode for team workflows.
Designing tool and resource interfaces for backend system integration.
JSON schemas, few-shot examples, and reliable extraction patterns.
Long documents, multi-turn conversations, multi-agent handoffs.
Code review, test generation, and PR feedback in automated pipelines.
Error handling, human-in-the-loop, self-evaluation patterns.
Five domains, weighted by their share of the scored content. Architecture and Claude Code workflows together account for nearly half the exam.
Every exam draws four scenarios at random from this set. Each scenario frames a group of questions that test architectural judgment in a realistic deployment.
An Agent SDK agent handles returns, billing disputes, and account issues via custom MCP tools (get_customer, lookup_order, process_refund, escalate_to_human). Target: 80%+ first-contact resolution while knowing when to escalate.
Claude Code accelerates day-to-day development — generation, refactoring, debugging, documentation. Integrate with custom slash commands, CLAUDE.md, and choose between plan mode and direct execution.
A coordinator delegates to specialized subagents — web search, document analysis, synthesis, and report generation — to produce comprehensive, cited reports.
Agents help engineers explore unfamiliar codebases, understand legacy systems, and automate repetitive tasks using Read, Write, Bash, Grep, Glob, and MCP servers.
Claude Code runs automated reviews, generates test cases, and gives PR feedback. Prompts must produce actionable feedback while minimizing false positives.
Extract information from unstructured documents, validate with JSON schemas, and integrate cleanly with downstream systems while handling edge cases gracefully.
Open any domain for practice questions, mental models, study notes, infographics, and a mind map.
Designing systems where Claude reasons, plans, and acts.
Building tools and MCP servers Claude can use safely.
Configuring Claude Code for real engineering work.
Eliciting precise, structured behavior from Claude.
Keeping Claude grounded across long, complex tasks.
Production data shows that in 12% of cases, your agent skips get_customer entirely and calls lookup_order using only the customer's stated name, occasionally leading to misidentified accounts and incorrect refunds. What change would most effectively address this reliability issue?
lookup_order and process_refund calls until get_customer has returned a verified customer ID.get_customer is mandatory before any order operations.get_customer first, even when customers volunteer order details.When a specific tool sequence is required for critical business logic — like verifying customer identity before processing refunds — programmatic enforcement provides deterministic guarantees that prompt-based approaches cannot. Options B and C rely on probabilistic LLM compliance, which is insufficient when errors have financial consequences. Option D addresses tool availability rather than tool ordering, which is not the actual problem.
Start with the domain that carries the most weight, or work through them in order.