Claude Academy Login: Access, Courses, and CCA-F Prep

Searching for the claude academy login and wondering what you actually get once you are inside? Anthropic Academy is a free learning portal, delivered on the Skilljar platform, that sits inside the Claude Partner Network. Access is open to anyone who joins the Partner Network at no cost. This guide explains how to get in, what the courses cover, and where the Academy ends and the actual CCA-F certification exam begins.

How do you access Anthropic Academy and log in?

Access follows a two-step path. First, apply to the Claude Partner Network on anthropic.com. Membership is free and, as of 3 June 2026, more than 40,000 firms have applied. Once your application is accepted, Anthropic provisions an Academy account on Skilljar and sends a welcome email with a login link. You authenticate with the email address used during Partner Network registration. There is no separate Academy registration form.

If you already have a Partner Network account and cannot find the login link, check your original welcome email or navigate directly to the Anthropic Academy subdomain on Skilljar and use "forgot password" with your Partner Network email. Anthropic does not publish a persistent public URL for the Academy portal in its main documentation, so the welcome email link is the canonical entry point.

What courses does Anthropic Academy offer, and are they free?

All Academy courses are free to enrolled partners. The catalogue covers the full Claude stack, from API fundamentals through agentic system design and the Model Context Protocol. Courses are self-paced video modules with knowledge checks, and Anthropic issues a completion certificate for each finished course. Those certificates are distinct from the CCA-F credential; more on that distinction below.

The table below maps the major course themes to the five CCA-F exam domains and their weightings, per Anthropic's exam guide:

Domain 1 carries the heaviest single weight at 27%, which is why the agentic architecture modules deserve the most study time. Our Agentic Architecture & Orchestration concept library maps every task statement in that domain to atomic practice questions.

What is the difference between an Academy completion certificate and the CCA-F?

This is the question we see most often from candidates who have finished all the Academy modules and assume they are now certified. They are not, and the distinction matters.

An Academy completion certificate confirms that you watched and passed the knowledge checks for a given course. It carries no proctored assessment, no scaled score, and no industry-recognised credential status. It is a learning record, not a professional certification.

The Claude Certified Architect, Foundations (CCA-F) exam is Anthropic's first professional certification, launched 12 March 2026. It costs $99 per attempt, runs 60 scenario-based multiple-choice questions, and requires a scaled score of 720 out of 1000 to pass. The exam is delivered online-proctored or at a test centre. Passing it earns the CCA-F credential, which is the qualification that appears in the Claude Partner Network's certified-individual count (10,000+ as of 3 June 2026).

In short: Academy courses prepare you; the CCA-F exam certifies you. Do not skip the exam and assume the certificates are equivalent.

How do the Academy courses map to the hardest CCA-F domains?

The two domains that trip up the most candidates are Domain 1 (Agentic Architecture, 27%) and the pair of Domain 3 and Domain 4 (Claude Code and Prompt Engineering, 20% each). Here is how Academy content aligns with each.

Domain 1: Agentic Architecture (27%)

Academy modules on multi-agent orchestration cover hub-and-spoke topologies, coordinator responsibilities, and subagent spawning. The exam tests these patterns in scenario form: given a broken pipeline, identify the root cause and the proportionate fix. Concepts like hub-and-spoke architecture and parallel subagent spawning are tested directly. The Academy modules give you the vocabulary; you still need to practise applying it under scenario pressure.

Domain 2: Tool Design & MCP Integration (18%)

MCP modules in the Academy walk through server configuration, scoping hierarchy, and the isError flag pattern. The exam rewards candidates who can distinguish an access failure from a valid empty result and who know when to split a tool versus when to rewrite its description. Our Tool Design & MCP Integration concept library covers all 30 task statements in this domain.

Domain 3: Claude Code Configuration (20%)

The Claude Code modules cover the three-level configuration hierarchy (user, project, and repository), CLAUDE.md authoring, and CI/CD integration. The exam consistently rewards deterministic, root-cause solutions over probabilistic ones. If a scenario describes a misconfigured hook or a missing environment variable, the correct answer traces the failure to its source rather than adding a retry wrapper.

Domain 4: Prompt Engineering & Structured Output (20%)

Academy prompt engineering content covers XML tagging, few-shot construction, and schema design. The exam tests whether you can identify why a prompt produces inconsistent output and select the minimum-change fix. Our Prompt Engineering & Structured Output concept library breaks each technique into a decision rule you can apply in under 90 seconds per question.

Do you need hands-on experience before the exam?

Anthropic's guidance for the CCA-F positions it as an architect-level credential. The exam scenarios assume familiarity with the Messages API, tool-use payloads, and agentic loop mechanics. Candidates who have only watched Academy videos without building anything tend to struggle with the scenario-based format because the questions describe real failure modes, not textbook definitions.

A practical minimum is to have built at least one agentic system that touches the Messages API, one MCP server integration, and one structured-output pipeline before sitting the exam. That gives you the pattern-recognition needed to read a scenario and immediately identify whether the failure is a stop_reason field misinterpretation, a tool description conflict, or a context window problem.

You do not need a computer science degree or published research. The exam tests applied architecture judgment, not academic credentials.

How do token economics affect what the exam tests?

Cost and reliability are not separate concerns in the CCA-F; they are the same concern. Several Domain 5 scenarios describe a production system that is either running over budget or producing inconsistent outputs, and the correct answer involves a context management technique rather than a model upgrade.

Key mechanics the exam tests include:

1. Prompt caching: Anthropic's prompt caching documentation describes how repeated system prompt prefixes can be cached to reduce input token costs. The exam tests when caching applies and when it does not.
2. Asynchronous message batches: The Message Batches API allows up to 10,000 requests per batch with a 29-day result expiry window, per Anthropic's Batches API documentation. The exam distinguishes scenarios where batch processing is appropriate from those requiring synchronous responses.
3. Context window management: Scenarios describe agents whose performance degrades over a long session. The correct fix is usually a summary injection for fresh sessions rather than increasing the model tier.

The exam does not ask you to calculate token costs numerically. It asks you to select the architecturally correct response to a cost or reliability symptom.

When should you use Opus versus Sonnet, and does the exam test this?

Model selection appears in Domain 1 and Domain 4 scenarios. The exam's consistent principle is proportionate tool for the task: use a more capable model only when the task genuinely requires it, and prefer deterministic enforcement over probabilistic model judgment for high-stakes decisions.

A common scenario pattern: a coordinator agent is using the most capable model for every subagent call, including trivial data-formatting steps. The correct answer routes formatting tasks to a lighter model and reserves the heavyweight model for reasoning-intensive steps. This is the model-driven vs pre-configured decision-making concept in practice.

The Academy courses introduce model tiers. The exam tests whether you can apply the selection logic under scenario pressure.

How does AI Skill Certs complement Anthropic Academy?

Anthropic Academy provides the conceptual foundation. AI Skill Certs (this platform) provides the scenario-based practice layer that the Academy does not offer.

Our practice exams are 60 questions, scored on the same 100-to-1000 scale as the real exam, with 720 as the passing bar. The adaptive engine uses Bayesian Knowledge Tracing with a 0.90 mastery threshold, which means it keeps routing you to your weak domains until you have genuinely consolidated them, not just seen them once.

The concept library at /concepts covers 174 atomic concepts mapped to all five domains and all 30 task statements. Archie, our Socratic tutor, never gives you the answer directly; it guides you with graduated hints so that you build the reasoning pattern rather than memorising a response.

AI Skill Certs is independent of Anthropic. We are not affiliated with, endorsed by, or approved by Anthropic. We tell you that plainly because it matters: our content is built from the published exam guide and official Anthropic documentation, not from any insider access.

The two resources are complementary. Watch the Academy modules to build vocabulary, then use AI Skill Certs to stress-test your judgment under exam conditions.

What is the recommended study sequence?

1. Join the Claude Partner Network and complete the claude academy login to access your course catalogue.
2. Work through Academy modules in domain-weight order: Domain 1 first (27%), then Domains 3 and 4 (20% each), then Domain 2 (18%), then Domain 5 (15%).
3. After each Academy module, read the corresponding concept pages on AI Skill Certs to see how the topic maps to specific task statements.
4. Run a diagnostic practice exam on AI Skill Certs to identify your weakest domains before you have invested significant study time.
5. Use Archie to work through the concepts where your diagnostic score is below 720.
6. Run a full 60-question practice exam. If you score above 720 consistently across two consecutive attempts, book the real exam at $99.
7. On exam day, read each scenario for the failure symptom first, identify the root cause, then select the proportionate fix. That heuristic covers the majority of CCA-F question patterns.