Exam orientation: format, scenarios, scoring, registration
Fifteen minutes that stop you studying the wrong way: how the CCA-F actually asks questions, how it is scored, what the proctor allows, and how to plan four weeks of preparation.
0.1 What the CCA-F is — and who should sit it
The Claude Certified Architect – Foundations (CCA-F) is Anthropic's first official certification, launched 12 March 2026 on Anthropic Academy. It validates that you can make informed architectural trade-offs when building real systems with the Claude API, Claude Code, the Claude Agent SDK, and the Model Context Protocol (MCP) — not that you can recite documentation.
Anthropic's stated candidate profile is roughly six or more months of hands-on building across those four surfaces. If you are still learning API fundamentals or mostly use Claude through the chat interface, work through this course first and sit the exam when the readiness exam says you are consistently above the pass line — the exam fee is real money and retakes cost time.
Access note: the exam launched gated to Claude Partner Network members (free to join), with wider public access expected during 2026 — still gated as of this writing. Registration is a two-step flow: eligibility/purchase through Anthropic's Partner Academy, then scheduling through Pearson VUE (see 0.2). Check the current status on Anthropic Academy before planning a date.
0.2 Format, scoring and proctoring
| Questions | 60 multiple-choice: one correct answer, three distractors. No labs, no free text. |
| Time | 120 minutes — an average of 2 minutes per question. Non-pausable. |
| Scoring | Scaled score of 100–1000; 720 passes. Scaled means you cannot compute "X of 60 correct = pass" exactly — aim well above the line in practice. |
| Fee | $125 listed on the official certification page (partner pricing has differed). |
| Proctoring | Delivered via Pearson VUE (OnVUE online proctoring, or an in-person test center) — closed book, no AI assistance. This moved off the exam's original ProctorFree setup in a June/July 2026 migration, alongside the exam's formal code becoming CCAR-F (still commonly called CCA-F). |
| Retakes | Up to 4 attempts per rolling 12 months, with escalating waits: 14 days after attempt 1, 30 after attempt 2, 90 after attempt 3. Recertification cycle is 12 months. |
Practical consequences: with 2 minutes per question, you cannot afford to reason every option from first principles. The candidates who finish comfortably are the ones who recognise the seven anti-patterns on sight and eliminate two options instantly.
0.3 The six scenarios — your exam is four of them
Every question on the CCA-F is anchored to a business scenario. The pool holds six; each sitting draws four at random, and all 60 questions are framed inside those four contexts. You cannot pick, so you prepare all six:
| Scenario | What it really tests |
|---|---|
| Customer Support Resolution Agent | Escalation design, tool error recovery, session state across long conversations. |
| Code Generation with Claude Code | Team-wide configuration: CLAUDE.md hierarchy, slash commands, plan mode vs direct execution. |
| Multi-Agent Research System | Coordinator/subagent orchestration, context passing, graceful failure handling. |
| Developer Productivity | Built-in tools (Read, Grep, Glob, Bash), MCP integration for codebase navigation. |
| Claude Code in CI/CD | Headless -p mode, --output-format json, session isolation, minimising false positives. |
| Structured Data Extraction | JSON schemas, nullable fields against hallucination, validation loops, Batch API trade-offs. |
Lesson 6 walks each scenario in depth; our readiness exam tags every question with its scenario so you feel the format early.
0.4 How questions are built (and how to eliminate)
CCA-F distractors are not random wrong answers — they are plausible architectural mistakes, drawn from a small set of failure modes: enforcing rules through prompts instead of hooks, trusting model self-reported confidence, using the Batch API in real-time flows, growing the context window instead of restructuring, silent subagent failures, unrestricted tool access, and flat agent topologies.
That gives you a repeatable elimination method: read the stem, identify the constraint word ("must never", "real-time", "cost-optimal", "20,000 documents"), and strike every option that matches a known anti-pattern. Usually exactly one option survives. Every quiz explanation in this course names the anti-pattern behind each wrong option to train that reflex.
0.5 Registration and the official resources
You register through Anthropic's Partner Academy for eligibility and purchase, then schedule the sitting itself through Pearson VUE. The certification page carries three documents worth reading before exam day: the official Exam Guide (includes sample questions), the Certification Terms, and the Exam Policy. Anthropic's full-length official practice exam was retired in the same mid-2026 migration to Pearson VUE — only the Exam Guide's sample questions remain official. Our readiness exam and the checkpoint quizzes in every lesson are the closest full-length substitute; run through this course and aim for a comfortable, repeated margin above 720 before booking.
Anthropic's free self-paced courses (Claude 101, Building with the Claude API, the MCP courses, Claude Code in Action, subagents and skills) are the official syllabus. This course compresses and cross-links that material into the exam's own structure — it replaces none of the hands-on practice, which is where the exam is actually won.
0.6 How to use this course — a 4-week plan
| Week | Study | Build (non-negotiable) |
|---|---|---|
| 1 | Lesson 1 (Domain 1 — 27%, the biggest) | Write a bounded agentic loop with real tool calling and error results. |
| 2 | Lessons 2 + 3 (MCP, Claude Code) | Build one small MCP server; set up CLAUDE.md + rules + one hook in a real repo. |
| 3 | Lessons 4 + 5 (structured output, context) | Ship a schema-validated extraction pipeline with a retry loop. |
| 4 | Lessons 6 + 7 (scenarios, anti-patterns), readiness exam twice | Fix your weakest domain from the diagnosis; retake the readiness exam until consistently above 720; book. |
Progress is saved in your browser — the rings on the course hub track each lesson. Start with Domain 1.
Quick glossary
Every term below appears somewhere in this course. If a question uses a word you're unsure of, check here before re-reading the whole lesson.
Agentic loop
The repeating cycle behind any Claude-powered agent: send a request, check what Claude wants to do next (stop_reason), run a tool if asked, send the result back, repeat until Claude is done.
stop_reason
A field on Claude's response telling your code why it stopped talking — e.g. it wants to call a tool (tool_use), it finished (end_turn), or it ran out of room (max_tokens).
tool_use / tool_result
tool_use is Claude asking your code to run something (a search, a database lookup...). tool_result is your code's answer, sent back so Claude can continue.
Subagent / Agent tool
A helper Claude spawns to do one focused piece of work in its own separate memory space, then reports back a summary — so the main conversation doesn't get cluttered with the helper's scratch work.
Hooks (PreToolUse / PostToolUse)
Small scripts that run automatically around a tool call — one right before it (can block it entirely) and one right after (can react to the result). Used for rules that must always hold, not just usually hold.
MCP (Model Context Protocol)
A standard way to connect Claude to external systems — databases, internal tools, company APIs — through a shared plug-in format instead of custom one-off code every time.
MCP Tools vs Resources vs Prompts
The three things an MCP connection can offer: Tools are actions Claude decides to call itself; Resources are data the surrounding app decides to hand Claude; Prompts are ready-made instructions a person picks, like a menu option.
.mcp.json
A config file listing which external MCP connections a project uses — shared with the whole team via version control, with any secret keys kept out of it and read from environment variables instead.
CLAUDE.md
A plain-text file Claude reads automatically at the start of every session in a project — the place for standing instructions like coding conventions or how the codebase is organized.
.claude/rules/
Instructions that only switch on for certain files (e.g. only when editing test files), instead of loading into every single session the way CLAUDE.md does.
Agent Skill
A packaged set of instructions Claude loads by itself when a task matches, without anyone typing a command — think of it as an instruction manual Claude picks up automatically off the shelf.
allowed-tools vs disallowed-tools
Two different settings that sound alike: allowed-tools just skips the usual 'are you sure?' prompt for tools you've pre-approved — it doesn't block anything else. disallowed-tools is what actually removes a tool from what's available.
Plan mode
A mode where Claude describes what it's about to do and waits for a yes before touching any files — used for bigger or riskier changes, skipped for small obvious ones.
Message Batches API
A cheaper way to send Claude a large pile of requests when nobody is waiting on the answer right away — costs about half as much, but can take up to a day to come back.
Prompt caching
A way to avoid Claude re-reading the same large, unchanging block of text (like a system prompt or a codebase) on every single request — it's remembered for a while, which is both cheaper and faster.
Context window
The total amount of conversation and text Claude can 'see' at once for a given request. When a conversation runs long, older parts may need to be summarized or dropped to make room.
Hub-and-spoke (multi-agent)
A setup where one coordinator agent hands out tasks to several helper agents and collects their answers — the helpers never talk directly to each other, only to the coordinator.
Anti-pattern
A design choice that looks reasonable but causes real problems in practice — this course names seven specific ones the exam likes to hide inside wrong answers.
Unofficial study resource — not affiliated with, endorsed by, or sponsored by Anthropic. Exam logistics reflect Anthropic's published information as of July 2026; always confirm current details on Anthropic Academy before booking.