Anthropic Free Courses: CCA-F Exam Prep Mapped

If you are preparing for the Claude Certified Architect, Foundations exam (CCA-F), the first question most candidates ask is whether Anthropic free courses cover enough ground to pass. The short answer: they cover the conceptual layer well, but the exam's scenario-based questions demand deliberate practice beyond passive video watching. This guide maps every publicly listed Anthropic Academy course to the five CCA-F domains, flags the gaps, and tells you where to fill them.

What is Anthropic Academy and how do you access it?

Anthropic Academy is Anthropic's official self-paced learning platform, delivered through Skilljar. Access is free: visit academy.anthropic.com, create a Skilljar account (email registration, no payment required), and enrol in any course. Completion earns a Skilljar-issued certificate of completion, which is distinct from the CCA-F credential. The CCA-F is a separately proctored exam costing $99 per attempt, scored on a 100-to-1000 scale with a passing mark of 720.

The Academy launched alongside the CCA-F on 12 March 2026 as part of the Claude Partner Network, a $100M programme. As of 3 June 2026, more than 10,000 individuals hold the CCA-F certification, which signals that the free content alone is not a bottleneck for motivated candidates.

Which Anthropic Academy courses exist and what do they cover?

Anthropic has published a growing catalogue. The table below maps the publicly listed courses to the five CCA-F exam domains and their weightings, so you can prioritise by exam impact rather than by course order.

The catalogue has grown from its original 13-course framing to 17 publicly listed courses as of May 2026. Anthropic has not designated any subset as "mandatory" for the CCA-F; the exam guide references the five domains and 30 task statements, not specific Academy courses.

Which courses should solution architects prioritise first?

Domain 1 (Agentic Architecture & Orchestration) carries the heaviest weight at 27%, so it earns first priority. Start with Introduction to Claude API to understand the Messages API request-response cycle, then move to Introduction to Subagents for delegation patterns. Our Agentic Architecture & Orchestration concept library covers 30-plus atomic concepts that go deeper than the Academy videos, including hub-and-spoke architecture and parallel subagent spawning.

Domain 3 (Claude Code Configuration & Workflows) and Domain 4 (Prompt Engineering & Structured Output) are each weighted at 20%, making them joint second priority. Introduction to Claude Code and Introduction to Claude Cowork address real-world file workflows and the three-level configuration hierarchy that the exam tests directly. Prompt Engineering with Claude and Structured Output with Claude cover the output-shaping techniques that appear across multiple domain scenarios.

Domain 2 (Tool Design & MCP Integration) at 18% is well served by Introduction to MCP and Tool Use and Function Calling. Pay particular attention to the isError flag pattern and tool description design, both of which appear in exam distractors. See our Tool Design & MCP Integration concepts for the full breakdown.

Domain 5 (Context Management & Reliability) at 15% is the lightest domain by weight but not by difficulty. Context and Memory Management and Prompt Caching with Claude are the relevant Academy courses. The prompt caching course covers the token minimums that candidates frequently misremember under exam conditions.

Do non-technical architects need AI Fluency or Claude 101 first?

Yes, with a caveat. AI Fluency and Claude 101 are designed for business stakeholders rather than builders. If you are comfortable reading Python and JSON, you can skip both and begin with Introduction to Claude API. If you are not, completing Claude 101 first will prevent you from getting lost in the technical courses. The technical courses assume you can read a JSON payload and understand basic HTTP request-response patterns; they do not assume prior ML knowledge.

Prerequisites in practice:

1. Comfortable with Python syntax and JSON structure: start at Introduction to Claude API.
2. Familiar with AI concepts but not coding: complete Claude 101, then Introduction to Claude API.
3. New to AI entirely: complete AI Fluency, then Claude 101, then the technical sequence.

How do the cloud-platform courses fit the exam?

Claude on AWS Bedrock and Claude on Google Cloud Vertex AI are not required for the CCA-F, but they are useful for candidates who will architect Claude deployments on those platforms. The exam tests Claude's native API behaviour and MCP integration patterns; it does not test cloud-provider-specific console navigation. That said, the Bedrock and Vertex AI courses reinforce Domain 1 and Domain 2 concepts through a different lens, which can help candidates who learn better from applied examples.

What do the Academy courses not cover that the exam tests?

The Academy courses are conceptual and demonstration-focused. The CCA-F delivers 60 scenario-based multiple-choice questions, each with one correct answer and three plausible distractors. Passive video consumption does not build the discrimination skill needed to reject a plausible-but-wrong option under time pressure.

Specific gaps we observe between Academy content and exam task statements:

• Agentic loop anti-patterns: the Academy introduces subagents but does not drill agentic loop anti-patterns at the depth the exam requires.
• Structured handoff to human agents: the exam tests when to escalate versus automate; the Academy covers this lightly.
• Hook pipeline architecture: PostToolUse hooks and tool call interception are exam topics not prominently featured in current Academy videos.
• Extended Thinking block structures: the Extended Thinking course introduces the concept, but the exam tests the precise structure of thinking blocks and when Extended Thinking is contraindicated.

Our Prompt Engineering & Structured Output and Context Management & Reliability concept libraries address these gaps with atomic, exam-mapped explanations.

How does an Anthropic Academy certificate compare to the CCA-F credential?

These are different artefacts with different market signals. An Anthropic Academy completion certificate confirms you watched and passed the in-course quizzes for a specific course. It is issued by Skilljar and carries no proctoring or identity verification.

The CCA-F is a proctored, scored credential tied to the Claude Partner Network. It is the credential that partner firms reference when staffing Claude-focused engagements. As of 3 June 2026, more than 40,000 firms have applied to the Claude Partner Network, creating genuine demand for verified CCA-F holders.

For solution architects offering enterprise consulting, the CCA-F is the credential that matters. The Academy certificates are useful for demonstrating breadth of study, not depth of competence.

How should you combine Academy courses with structured practice?

We recommend a three-phase approach:

1. Foundation phase (weeks 1 to 2): Complete the Academy courses mapped to your two weakest domains. Use the courses to build vocabulary and mental models, not to memorise answers.
2. Concept phase (weeks 3 to 4): Work through the atomic concepts in our concept library for all five domains. Each concept maps to specific task statements from the exam guide, so you can track coverage precisely. Our adaptive engine uses Bayesian Knowledge Tracing with a 0.90 mastery threshold to surface the concepts you have not yet consolidated.
3. Practice phase (weeks 5 to 6): Sit full 60-question practice exams scored on the 100-to-1000 scale. Review every distractor you chose incorrectly; the exam rewards root-cause tracing over pattern matching.

AI Skill Certs is an independent prep platform; we are not affiliated with or endorsed by Anthropic. We build on the same public exam guide that every candidate can access.

Are there prerequisites for the CCA-F exam itself?

Anthropic does not publish formal prerequisites for the CCA-F. There is no required course sequence, no mandatory Academy completion, and no minimum experience threshold stated in the exam guide. In practice, candidates without hands-on Claude API experience find the scenario-based questions significantly harder, because the distractors are designed to catch candidates who understand concepts abstractly but have not applied them.

The exam consistently rewards deterministic solutions over probabilistic ones when stakes are high, proportionate fixes over over-engineered responses, and root-cause tracing over symptomatic remediation. These are skills built through practice, not through passive learning.