Claude Certified Architect — Foundations (CCA-F)

A study repository for the Anthropic Claude Certified Architect Foundations certification exam.

Exam At a Glance

Detail	Info
Format	60 scenario-based multiple-choice questions
Duration	120 minutes (no breaks, no AI assistance)
Platform	ProctorFree (proctored)
Passing Score	720 / 1,000
Results	Within 2 business days (domain breakdown included)
Cost	Free (first 5,000 partner employees), $99/attempt thereafter

Exam Domains

#	Domain	Weight
1	Agentic Architecture & Orchestration	27%
2	Claude Code Configuration & Workflows	20%
3	Prompt Engineering & Structured Output	20%
4	Tool Design & MCP Integration	18%
5	Context Management & Reliability	15%

Domains 1 + 2 = 47% of the exam. Treat this as a systems design exam, not a prompting fundamentals test.

Exam Scenarios (4 of 6 randomly selected)

1. Customer Support Resolution Agent
2. Code Generation with Claude Code
3. Multi-Agent Research System
4. Developer Productivity with Claude
5. Claude Code for CI/CD
6. Structured Data Extraction

---

Actionable Study Plan

Week 1 — Foundations (6–8 hrs)

Complete the four core Anthropic Academy courses in order:

• Claude 101 — Claude fundamentals
• Building Applications with the Claude API — API integration, streaming, auth, structured output
• Introduction to Model Context Protocol — MCP protocol and tool design
• Claude Code in Action — CLAUDE.md, workflows, skills, CI/CD

Download the Official Exam Guide PDF from Skilljar and work through all 12 sample questions.

---

Week 2 — Domain 1: Agentic Architecture (6–8 hrs)

27% of the exam — highest value domain.

• Read Agent SDK Overview — agentic loop mechanics and subagent patterns
• Read Building Agents with the Claude Agent SDK — hooks, orchestration, sessions
• Clone and run Agent SDK Python examples — hooks, custom tools, fork_session
• Paste the Domain 1 study prompt (in STUDY-GUIDE.md) into Claude and complete all 7 task statements
• Build: a coordinator agent with 2 subagents, explicit context passing, a programmatic prerequisite gate, and a PostToolUse normalisation hook
• Score 8+/10 on the Domain 1 practice questions before moving on

---

Week 3 — Domains 2, 3 & 4: Claude Code + Prompting + Tools (6–8 hrs)

Domain 2 — Claude Code Configuration (20%)

• Read Claude Code official docs — CLAUDE.md hierarchy, rules directory, slash commands, skills
• Read Claude Code CLI reference — all flags including -p, --output-format json, --allowedTools
• Paste the Domain 2 study prompt into Claude and complete all 6 task statements
• Build: project with CLAUDE.md hierarchy, .claude/rules/ with glob patterns, a skill using context: fork, and a CI script using -p flag

Domain 3 — Prompt Engineering & Structured Output (20%)

• Read Anthropic Prompt Engineering docs
• Read API Tool Use documentation — tool_use, tool_choice config, JSON schema enforcement
• Paste the Domain 3 study prompt into Claude and complete all 6 task statements
• Build: an extraction pipeline using tool_use with required/optional/nullable fields, a validation-retry loop, and a Batch API run

Domain 4 — Tool Design & MCP Integration (18%)

• Read MCP Integration for Claude Code
• Browse MCP specification and community servers
• Read Agent SDK TypeScript repo — tool definition patterns and structured error responses
• Paste the Domain 4 study prompt into Claude and complete all 5 task statements
• Build: two MCP tools with intentionally similar/ambiguous descriptions, then fix them — experience the before/after routing difference

---

Week 4 — Domain 5 + Practice Tests (4–6 hrs)

Domain 5 — Context Management & Reliability (15%)

• Read Building Agents with the Claude Agent SDK — context management, error propagation, escalation
• Read Agent SDK session docs — --resume, fork_session, /compact
• Paste the Domain 5 study prompt into Claude and complete all 6 task statements
• Build: coordinator with 2 subagents, persistent case facts block, simulated timeout with structured error propagation, conflicting sources with attribution

Practice exams

• Complete the official 60-question practice test (available after registration)
• Complete claudecertifications.com free practice questions
• Udemy — Anthropic Claude Certified Architect: 3 Full Practice Exams — 3 × 60-question practice exams with explanations
• Target: 900+ / 1,000 before booking the real exam

---

Practice Score Benchmark

Score	Readiness
Below 700	Not ready — revisit weak domains
700–799	Borderline — more scenario practice needed
800–899	Good — review all missed questions
900+	Ready to book the real exam

---

Workflow: after each practice exam

After every practice exam, run the log-exam-attempt skill to update your two personal study artifacts:

• PERFORMANCE-IMPROVEMENT-PLAN.md — cumulative trend analysis: progression table, rules derived from your misses, decision tree, per-attempt breakdown.
• MISSED-QUESTIONS-RETAKE.md — retake bank of missed questions with an answer key at the bottom you can cover while drilling.

Both files are gitignored. They are personal and should never be committed.

How it works

Invoke the skill with your exam results:

/log-exam-attempt <path-to-results-file>

Or paste the results directly in chat and mention the skill. Minimum required info per miss: domain, what you picked, what was correct. Full question text and options are ideal but not required (the skill falls back to flashcard format when only excerpts are available).

The skill is a coordinator that spawns two subagents in parallel (hub-and-spoke, the same pattern Domain 1 tests on):

• improvement-plan-updater — appends a new attempt section, updates the progression table, derives new rules, flags repeat misses, revises study priorities.
• retake-bank-updater — adds missed questions to Part A (full text) or Part B (flashcards), extends the answer key.

On first run, both files are created from the templates in .claude/templates/.

Migrating hand-crafted artifacts

If you already have personal PERFORMANCE-IMPROVEMENT-PLAN.md or MISSED-QUESTIONS-RETAKE.md files you maintained by hand, stash them first so the skill starts from the template:

mv PERFORMANCE-IMPROVEMENT-PLAN.md PERFORMANCE-IMPROVEMENT-PLAN.stash.md
mv MISSED-QUESTIONS-RETAKE.md MISSED-QUESTIONS-RETAKE.stash.md

*.stash.md is gitignored, so your history stays local. You can reconcile the stash into the new structure at your own pace.

---

Repository Structure

claude-code-certification/
├── README.md                                          # Exam overview and actionable study plan
├── STUDY-GUIDE.md                                     # Per-domain notes, exam traps, and study prompts
├── .gitignore                                         # Keeps personal study artifacts out of git
└── .claude/
    ├── skills/
    │   └── log-exam-attempt/SKILL.md                  # Post-exam coordinator skill
    ├── agents/
    │   ├── improvement-plan-updater.md                # Subagent: maintains PERFORMANCE-IMPROVEMENT-PLAN.md
    │   └── retake-bank-updater.md                     # Subagent: maintains MISSED-QUESTIONS-RETAKE.md
    └── templates/
        ├── PERFORMANCE-IMPROVEMENT-PLAN.template.md   # Seed for first-run users
        └── MISSED-QUESTIONS-RETAKE.template.md        # Seed for first-run users

Personal (gitignored) files that appear at the repo root after your first /log-exam-attempt run:

PERFORMANCE-IMPROVEMENT-PLAN.md    # your trend analysis, grows per attempt
MISSED-QUESTIONS-RETAKE.md         # your retake bank, grows per attempt