Unofficial study guide for the Claude Certified Architect — Foundations (CCAR-F) certification exam by Anthropic.
This guide covers all 5 exam domains with detailed explanations, code examples, anti-patterns, decision frameworks, and practice questions.
It is actively maintained and updated regularly to track changes to Anthropic's models, APIs, and Claude Code. Last refresh: September 2026 — see CHANGELOG.md for what changed and when.
Anthropic now runs four certifications. This repo's deep-dive guide covers CCAR-F; overview guides for the other three (sourced from the official v1.0 exam guides) live in certs/:
| Code | Credential | Audience | Fee | Guide |
|---|---|---|---|---|
| CCAO-F | Associate — Foundations | Business / productivity users (non-developer) | $99 | Overview |
| CCAR-F | Architect — Foundations | Solution architects | $125 | Full study guide — this repo's main guide |
| CCAR-P | Architect — Professional | Senior architects owning the full solution lifecycle | $175 | Overview |
| CCDV-F | Developer — Foundations | Engineers shipping Claude apps, agents, and workflows | $125 | Overview |
All four are 120-minute proctored exams delivered via Pearson VUE, passing score 720/1,000, credentials valid 12 months (Exam Guides v1.0, effective July 2026). Renewal is free if done on time: review what changed and pass a non-proctored assessment; a lapsed credential means sitting the full exam again. Registration needs a partner-domain email — personal addresses are rejected.
- Website: https://dnacenta.github.io/claude-certified-architect/
- PDF (English): https://dnacenta.github.io/claude-certified-architect/guide_en.pdf
The landing page and PDF are built and deployed automatically from this repo's markdown on every push to main (see .github/workflows/pages.yml). The PDF is generated from the source .md files, so it never drifts from the guide.
- Format: 60 multiple-choice, scenario-based questions in 120 minutes (proctored, closed-book)
- Passing score: 720/1000
- Scenarios: 4 of 6 randomly selected per exam
- Delivery: Pearson VUE (OnVUE online or test center), registered via the Anthropic Partner Academy
- Price / validity: $125 per attempt; certification valid 12 months, free on-time renewal via a non-proctored assessment (Exam Guide v1.0, effective July 2026)
- Target audience: Solution architects with 6+ months experience building with Claude APIs, Agent SDK, Claude Code, and MCP; access is gated to the Claude Partner Network
| Domain | Weight | Guide |
|---|---|---|
| Agentic Architecture & Orchestration | 27% | Domain 1 |
| Tool Design & MCP Integration | 18% | Domain 2 |
| Claude Code Configuration & Workflows | 20% | Domain 3 |
| Prompt Engineering & Structured Output | 20% | Domain 4 |
| Context Management & Reliability | 15% | Domain 5 |
See claude-certified-architect.md for the full overview including exam scenarios, anti-patterns, decision frameworks, a 4-week study plan, and official resources.
Code samples use claude-opus-5 — Anthropic's current default for complex agentic coding and enterprise work.
| Model | ID | Context | Max output | Price (in / out per MTok) |
|---|---|---|---|---|
| Claude Fable 5.1 | claude-fable-5-1 |
1M | 128k | $10 / $50 (cache reads $0.25) |
| Claude Opus 5 | claude-opus-5 |
1M | 128k | $5 / $25 |
| Claude Sonnet 5 | claude-sonnet-5 |
1M | 128k | $2 / $10 |
| Claude Haiku 4.5 | claude-haiku-4-5 |
200k | 64k | $1 / $5 |
Fable 5, Opus 4.8, 4.7, 4.6, 4.5, Sonnet 4.6, and Sonnet 4.5 are legacy but still available (Opus 4.1 retired 2026-08-05). Fable 5.1 (released 2026-09-01) rejects forced tool_choice and binds thinking blocks to the model and history that produced them — see the main guide.
- Official CCAR-F Exam Guide (PDF, v1.0) — the March 2026 launch guide is superseded
- Anthropic Partner Academy — CCAR-F registration · Prep courses · Certification FAQ
- Pearson VUE — Anthropic certification program
- Anthropic Skilljar — Building with Claude API
- 12-Week Training Program (GitHub)
- Building Effective Agents (Anthropic Research)
- Claude API docs · Models overview · Claude Code docs · Claude Agent SDK · MCP
The main guide's Resources section has the full, categorised link list.
Docs note: Anthropic split its documentation in July 2026. API docs live at
platform.claude.com/docs/en/*, Claude Code docs atcode.claude.com/docs/en/*; the olddocs.anthropic.com/en/docs/*URLs still redirect. The SDK was renamed from "Claude Code SDK" to the Claude Agent SDK.
This guide was inspired by and based on the exam breakdown by @hooeem on X. Full credit to them for compiling the domain coverage, scenarios, and key concepts from the official exam guide.
I'm also building open-source tools around persistent agents, structured memory, and Claude Code usage visibility — same problem space as the agentic / MCP / context domains in this guide, different angle:
- recall-echo — persistent, confidence-weighted memory for coding agents (MCP-friendly). Early but usable.
- pulse-null — single-binary scaffold for longer-running AI entities. Research prototype — expect bugs.
- pulse-limits — Claude, Codex and Grok plan limits as a retro patient monitor: macOS menu bar, Waybar, terminal. One Rust binary.
brew install pulse-null/tap/pulse-limits && pulse-limits install. No account, no server — it reuses the logins your CLIs already keep. - pulse-null.com — overview of the work.
Not affiliated with Anthropic or the Claude Certified Architect program. Solo project; stars and issues help.
Feel free to reach out: dnacenta@pulse-null.com
This is an unofficial community study guide. Claude Certified Architect is a certification program by Anthropic.