# Claude Certified Architect Prep

> Exam-prep platform for Anthropic's four Claude certifications: CCAR-F (also written CCA-F; flagship + deepest coverage), CCAR-P (CCA-P), CCDV-F (CCD-F), and CCAO-F (CCA-A). 71 technical-primitive concept pages across the four certs, 25 production scenarios, 13 Anthropic Academy course summaries, a free diagnostic + warm-up + full mock per cert, and adaptive study plans.

## Audience
- AI engineers & architects preparing for CCAR-F (CCA-F), CCAR-P (CCA-P), CCDV-F (CCD-F), or CCAO-F (CCA-A); all use a 720/1000 pass mark
- Architects building agentic systems with Claude (tool calling, MCP, hooks, escalation)
- Technical leads evaluating Claude vs GPT-5 / Gemini for production loops

## Freshness
- Last reviewed: 2026-09-12
- Refresh cadence: monthly for content, weekly for evidence ledger + blog

## Concepts (Pillar 2, 71 technical primitives across 4 certs)

### D1 · Agentic Architectures (27% of exam)
- [D1.1 Agentic Loops](https://claudearchitectcertification.com/concepts/agentic-loops.md), An agentic loop turns one API response into a controlled, multi-turn system.
- [D1.2 Subagents](https://claudearchitectcertification.com/concepts/subagents.md), Subagents are specialized, isolated agents spawned by a coordinator to handle domain-specific tasks while preserving conversation isolation.
- [D1.3 Stop Reason](https://claudearchitectcertification.com/concepts/stop-reason.md), stop_reason is the authoritative struct field that says why Claude stopped.
- [D1.4 Session State & Persistence](https://claudearchitectcertification.com/concepts/session-state.md), Session state is the running message list (and any external persistence) that gives an agentic loop continuity.
- [D1.5 Human-in-the-Loop Escalation](https://claudearchitectcertification.com/concepts/escalation.md), Escalation is the deterministic handoff path when policy thresholds, low confidence, or repeated failure conditions are hit.
- [D1.6 Subagent State Handoff](https://claudearchitectcertification.com/concepts/subagent-state-handoff.md), Subagents hand off state three ways: structured messages (default - typed JSON envelopes through the orchestrator), tool results (tightest sync chain when receiver is a tool), or shared files (size or binary fallback only).
- [D1.7 Agent Instruction Files (CLAUDE.md / AGENTS.md)](https://claudearchitectcertification.com/concepts/agent-instruction-files.md), CLAUDE.
- [D1.8 Task Decomposition Strategies](https://claudearchitectcertification.com/concepts/task-decomposition.md), Task decomposition is the up-front architectural choice between two regimes: fixed decomposition (subtasks known in advance, run as a sequential prompt chain) and dynamic decomposition (a lead LLM determines subtasks at runtime because they depend on what earlier steps discover).

### D2 · Tool Design + Integration (18% of exam)
- [D2.1 Tool Calling](https://claudearchitectcertification.com/concepts/tool-calling.md), Tool calling is how Claude decides to invoke external functions and pass structured arguments.
- [D2.2 Tool Choice](https://claudearchitectcertification.com/concepts/tool-choice.md), tool_choice is the parameter that controls whether Claude can decide to use a tool ("auto"), must call any tool ("any"), or must call a specific tool by name.
- [D2.3 Model Context Protocol](https://claudearchitectcertification.com/concepts/mcp.md), MCP is a communication standard that lets Claude access pre-built tools, resources, and prompts from specialized servers without you writing integration code.
- [D2.4 SDK Hooks (Pre/PostToolUse)](https://claudearchitectcertification.com/concepts/hooks.md), Hooks are deterministic code that runs before or after tool calls.
- [D2.5 Tool Evaluation & Testing](https://claudearchitectcertification.com/concepts/evaluation.md), Evaluation is how you measure tool-call correctness against ground-truth datasets.
- [D2.6 Built-In Tools vs Custom Tools vs Skills vs MCP](https://claudearchitectcertification.com/concepts/agentic-tool-surface-selection.md), Built-in tools, custom tools, Agent Skills, and MCP servers are complementary layers, not a power ranking — one agent often uses several at once.

### D3 · Claude Code Configuration & Workflows (20% of exam)
- [D3.1 CLAUDE.md Hierarchy](https://claudearchitectcertification.com/concepts/claude-md-hierarchy.md), CLAUDE.
- [D3.2 Plan Mode](https://claudearchitectcertification.com/concepts/plan-mode.md), Plan Mode forces Claude Code to draft and approve a multi-step plan before executing.
- [D3.3 Skills](https://claudearchitectcertification.com/concepts/skills.md), Skills are reusable, on-demand task workflows defined in markdown with YAML frontmatter.
- [D3.4 Checkpoints & Session Management](https://claudearchitectcertification.com/concepts/checkpoints.md), Checkpoints save and restore conversation state.

### D4 · Prompt Engineering (20% of exam)
- [D4.1 System Prompts & Instructions](https://claudearchitectcertification.com/concepts/system-prompts.md), System prompts establish role, constraints, format, and tool guidance.
- [D4.2 Prompt Caching](https://claudearchitectcertification.com/concepts/prompt-caching.md), Prompt caching reduces cost (~90%) on repeated context like long system prompts and tool definitions.
- [D4.3 Batch API](https://claudearchitectcertification.com/concepts/batch-api.md), Message Batches API: 50% discount for async, non-time-sensitive workloads.
- [D4.4 Structured Outputs](https://claudearchitectcertification.com/concepts/structured-outputs.md), Structured outputs guarantee Claude returns JSON matching a schema instead of natural language.
- [D4.5 Vision & Multimodal](https://claudearchitectcertification.com/concepts/vision-multimodal.md), Vision lets Claude process images alongside text.
- [D4.6 Streaming](https://claudearchitectcertification.com/concepts/streaming.md), Streaming returns tokens as they're generated for low-latency UX.
- [D4.7 Attention Engineering](https://claudearchitectcertification.com/concepts/attention-engineering.md), Attention engineering is the discipline of placing critical context where the model attends most strongly, high in the prompt, in the system message, or in repeated facts blocks.
- [D4.8 4D Framework](https://claudearchitectcertification.com/concepts/4d-framework.md), Delegation, Description, Discernment, Diligence, Anthropic's 4D framework for agent prompt design.
- [D4.9 Prompt Engineering Techniques](https://claudearchitectcertification.com/concepts/prompt-engineering-techniques.md), Prompt engineering is a craft of seven techniques that turn a brittle one-shot prompt into a production-grade contract: few-shot examples, iterative refinement against an eval suite, anti-fabrication schemas, structured templates, explicit do-don't constraints, output anchoring via tools, and test-driven prompt edits.
- [D4.10 Multi-Instance & Multi-Pass Review Architectures](https://claudearchitectcertification.com/concepts/multi-pass-review-architectures.md), A multi-pass review architecture runs more than one independent Claude pass over the same artifact and reconciles the results, instead of trusting a single evaluation - it covers two related patterns: the evaluator-optimizer loop (generate, then a separate evaluator critiques, repeat) and fleet-style parallel review (many independent reviewer instances, each verifying its own findings, reconciled before a human sees them).

### D5 · Context + Reliability (15% of exam)
- [D5.1 Context Window Management](https://claudearchitectcertification.com/concepts/context-window.md), Context window management is how you keep long conversations within limits without dropping critical facts.
- [D5.2 Case Facts Block](https://claudearchitectcertification.com/concepts/case-facts-block.md), A case-facts block is an immutable set of transactional data (customer ID, order details, refund amount, policy limits) included at the top of every prompt during a multi-turn conversation.

### CCA-P · Claude Certified Architect - Professional (17 concepts)
- [CCARP-D1.1 Translating Business Requirements into Claude Solutions](https://claudearchitectcertification.com/concepts/requirements-to-solution-design.md), Translating a business problem into a Claude solution means producing a build-ready requirements spec, not picking an architecture pattern.
- [CCARP-D1.2 End-to-End Architecture Design (Input to Feedback Loop)](https://claudearchitectcertification.com/concepts/end-to-end-architecture-design.md), An architect-level design task is never just "the prompt in the middle" - it is the full pipeline: how input arrives, how it is processed (augmented LLM, workflow, or agent loop), how output is delivered, and how the system learns from what happened.
- [CCARP-D1.3 Workflow vs Agent: Choosing the Right Architecture](https://claudearchitectcertification.com/concepts/workflow-vs-agent-architecture-patterns.md), Pick the cheapest tier that actually satisfies the task: an augmented LLM call for a single unaided prompt, a workflow when the steps are known in advance, an agent only when the model must dynamically decide its own path.
- [CCARP-D1.4 Aligning Solutions to Business Value Pillars](https://claudearchitectcertification.com/concepts/business-value-alignment.md), Every architectural choice - pattern, model tier, guardrail overhead - has to trace back to a named business-value pillar an executive sponsor recognizes: efficiency, transformation, productivity, cost, or performance SLAs.
- [CCARP-D3.1 Authentication & Access Control for Agent Systems](https://claudearchitectcertification.com/concepts/auth-and-access-control-for-agents.md), Agent authentication and authorization is a stack of controls, not one setting: which identity the process runs as, what credentials each tool call carries, and which actions need a human in the loop.
- [CCARP-D3.2 Accuracy-Latency-Cost Tradeoff Analysis](https://claudearchitectcertification.com/concepts/accuracy-latency-cost-tradeoffs.md), Every production Claude deployment sits on a capability/speed/cost triangle, and the exam trap is treating it as one dial ('use the biggest model') instead of independent levers - model tier, effort, thinking budget, prompt caching, and batching - that trade one axis for another *within* a single model choice, not just across the model family.
- [CCARP-D3.3 Observability & Monitoring for Agents at Scale](https://claudearchitectcertification.com/concepts/observability-and-monitoring-at-scale.md), Observability for a Claude agent fleet spans layers you must monitor separately: platform-level usage/cost accounting (Anthropic's Admin/Analytics APIs), Managed Agents session tracing (a Console Tracing view plus native span events, scoped to agents run on that platform), and application-level execution tracing for a self-orchestrated agent built directly against the Messages API, where Anthropic does not natively provide distributed tracing.
- [CCARP-D3.4 Progressive Discovery vs Monolithic Context Strategy](https://claudearchitectcertification.com/concepts/progressive-disclosure-vs-monolithic-context.md), As an agent's tool/skill/knowledge surface grows, an architect chooses between loading everything upfront (monolithic - predictable, but token-expensive) or exposing capabilities on demand (progressive discovery - token-efficient, but dependent on the model finding the right thing at the right time).
- [CCARP-D5.1 Guardrails & Safe Deployment Practices](https://claudearchitectcertification.com/concepts/guardrails-and-safety-controls.md), Guardrails are not one control but a layered system: model-level training behavior, input-level content screening, tool-level permission policies, process-level sandboxing, and human review - each layer stops a different failure class.
- [CCARP-D5.2 LLM System Risks & Failure Modes](https://claudearchitectcertification.com/concepts/llm-risks-and-failure-modes.md), Architects must reason about failure modes that don't exist in deterministic software: non-determinism, prompt injection (direct and indirect), long-context jailbreak attacks, and - unique to agentic systems with tool access - a model taking a harmful real-world action rather than just emitting bad text.
- [CCARP-D5.3 Regulatory Compliance for AI Systems (GDPR, HIPAA, FedRAMP)](https://claudearchitectcertification.com/concepts/regulatory-compliance-for-ai-systems.md), Compliance is a match-the-surface-to-the-regime exercise: an architect picks the deployment surface (API direct, AWS Bedrock, GCP Vertex, Microsoft Foundry, Claude for Government) and account tier the workload requires, then explicitly turns on the controls that are not on by default.
- [CCARP-D5.4 Ethical AI Considerations (Bias, Fairness, Transparency)](https://claudearchitectcertification.com/concepts/ethical-ai-considerations.md), Ethical AI for an architect is not a vendor guarantee, it is three published artifacts to consult: Claude's Constitution (intended values, the training-time "final authority"), per-model System Cards (disclosed bias and safety evaluation results), and the Transparency Hub (aggregated Model Reports plus voluntary commitments).
- [CCARP-D6.1 Structured Stakeholder Discovery](https://claudearchitectcertification.com/concepts/stakeholder-discovery-and-requirements.md), Discovery for a Claude-based solution has to surface things traditional SDLC discovery skips: whether the task even fits a nondeterministic system, what "good enough" output looks like, and which architecture pattern the problem needs - before any model or design decision gets made.
- [CCARP-D6.2 Communicating AI Decisions & Value to Stakeholders](https://claudearchitectcertification.com/concepts/communicating-ai-decisions-and-value-to-stakeholders.md), An architect's technical trade-offs (model tier, workflow vs agent, guardrail depth) are business decisions in disguise, and Anthropic names the exact vocabulary to present them in: Capabilities, Cost and Speed, and Safety.
- [CCARP-D6.3 Stakeholder Feedback Loops & SLA Alignment](https://claudearchitectcertification.com/concepts/stakeholder-feedback-and-sla-alignment.md), A Claude-based system's "SLA" is not one number - an architect has to keep at least three infrastructure-side concepts straight, plus a fourth for quality: observed historical uptime (a retrospective status-page record, e.
- [CCARP-D6.4 Architecture Documentation & Implementation Handoff](https://claudearchitectcertification.com/concepts/architecture-documentation-and-handoff.md), Handoff for a Claude-based architecture needs everything a traditional diagram needs (components, data flow, integration points) PLUS a set of GENERAL artifacts every Claude-based system needs regardless of build tooling: the eval suite that defines "working," prompt/config version history, an ADR-style decision record, and a runbook for known failure modes.
- [CCARP-D6.5 AI Solution Lifecycle Management](https://claudearchitectcertification.com/concepts/ai-solution-lifecycle-management.md), Deployment isn't the finish line - an architect owns a continuous loop of discovery, design, handoff, monitoring, and iteration, because model behavior and usage patterns keep drifting after launch.

### CCD-F · Claude Certified Developer - Foundations (10 concepts)
- [CCDVF-D1.1 Building Agents: Client SDK vs Agent SDK vs Managed Agents](https://claudearchitectcertification.com/concepts/agent-construction-with-claude-agent-sdk.md), Anthropic gives developers three layers for building a Claude agent, and they are NOT interchangeable names for the same thing: the Client SDK / Messages API (you write the tool loop yourself), the Claude Agent SDK (Claude drives the loop autonomously inside your own process, pip install claude-agent-sdk / npm install @anthropic-ai/claude-agent-sdk), and Claude Managed Agents (a hosted REST API where Anthropic runs the loop and the sandbox).
- [CCDVF-D1.2 Agent Design Patterns: Memory, Context Management & Third-Party Frameworks](https://claudearchitectcertification.com/concepts/agent-frameworks-and-design-patterns.md), An agent's execution mechanism is the tool-use loop: call a tool, read real environmental feedback ("ground truth"), repeat until the task converges - that loop, not any framework, is what makes a system an agent.
- [CCDVF-D2.1 Systems Life Cycle Management for Claude Applications](https://claudearchitectcertification.com/concepts/systems-life-cycle-management.md), Systems life cycle management is a general IT discipline - plan, develop, test, deploy, operate, maintain, retire - not an Anthropic invention; NIST's SDLC bulletin frames it as "the overall process of developing, implementing, and retiring information systems.
- [CCDVF-D2.2 Software Engineering Foundations for Claude Applications](https://claudearchitectcertification.com/concepts/software-engineering-foundations-for-claude-apps.md), The Claude Messages API is a plain REST API that accepts and returns JSON - POST /v1/messages with a stateless request/response contract - so the ordinary software-engineering discipline you already use for any HTTP service (JSON contracts, async I/O, version control, code review, disciplined refactoring) is exactly what a production Claude integration needs, not a special category of its own.
- [CCDVF-D2.3 Claude Application Design Across Interfaces](https://claudearchitectcertification.com/concepts/claude-application-design-across-interfaces.md), Claude is one model family reached through several host surfaces, claude.
- [CCDVF-D4.1 Debugging LLM System Failures](https://claudearchitectcertification.com/concepts/debugging-llm-system-failures.md), A Claude application fails in two structurally different places: the integration layer (the call never reached a valid model turn - bad auth, malformed request, rate limit) or model output (the call returned 200, but the content is wrong - refused, truncated, or malformed).
- [CCDVF-D5.1 Claude Model Selection & Tradeoffs](https://claudearchitectcertification.com/concepts/model-selection-tradeoffs.md), Match the model tier to the task, not the other way round: Haiku for high-volume, straightforward work at the lowest cost/latency; Sonnet for the best speed/intelligence balance; Opus for complex, long-horizon agentic work; Fable as the highest-capability, slowest tier for work none of the other three can cover.
- [CCDVF-D5.2 LLM Fundamentals: Tokens, Sampling & Non-Determinism](https://claudearchitectcertification.com/concepts/llm-fundamentals-tokens-and-sampling.md), Claude is autoregressive: it predicts one token at a time from a probability distribution conditioned on everything before it, then appends that token and repeats.
- [CCDVF-D5.3 Technical Fundamentals: SDKs, REST, and WebSockets](https://claudearchitectcertification.com/concepts/technical-fundamentals-sdks-and-websockets.md), For standard model inference, the Messages API (POST /v1/messages) is the primary endpoint, tools, structured outputs, and streaming are request parameters on that same endpoint, not separate APIs or protocols.
- [CCDVF-D7.1 AI Application Security: Prompt Injection & Untrusted Input](https://claudearchitectcertification.com/concepts/ai-application-security-prompt-injection.md), Prompt injection splits into two threat models with different defenses: direct injection / jailbreaks (your own user crafts a bypass) mitigated by input screening and hardened system prompts, and indirect injection (a trusted user, but untrusted third-party content Claude processes) mitigated primarily by least privilege and action authorization on the tool call itself, backed by tool-result structuring - not by trusting Claude's training alone, and not by content formatting acting as a security boundary.

### CCA-A · Claude Certified Associate - Foundations (14 concepts)
- [CCAOF-D1.1 Prompting Strategy by Task Type: Analysis, Research, Drafting, Brainstorming](https://claudearchitectcertification.com/concepts/task-type-prompting-strategies.md), "Be clear and direct" is Anthropic's base rule for every Claude prompt, but how you apply it changes with the type of work.
- [CCAOF-D2.1 Evaluating Claude Output for Accuracy & Completeness](https://claudearchitectcertification.com/concepts/evaluating-output-accuracy-completeness.md), Accuracy and completeness are two independent checks, not one: an output can be fully correct on every claim it includes and still fail by silently dropping items, or it can cover everything asked and still get a fact wrong.
- [CCAOF-D2.2 Identifying Hallucinations, Inconsistencies & Bias in AI Output](https://claudearchitectcertification.com/concepts/identifying-and-validating-ai-output-errors.md), Hallucination is Anthropic's own term for Claude generating text that's factually incorrect or inconsistent with the material it was given, and it's a known limitation, not a rare bug: "users should not rely on Claude as a singular source of truth" and should scrutinize high-stakes output carefully.
- [CCAOF-D2.3 Editing & Adapting Outputs for the Intended Audience](https://claudearchitectcertification.com/concepts/adapting-outputs-for-audience.md), A first Claude draft is rarely the version you hand to its actual reader.
- [CCAOF-D2.4 Choosing Output Formats: Artifacts vs Inline vs Structured Data](https://claudearchitectcertification.com/concepts/choosing-output-formats-artifacts-vs-inline.md), Choosing how a Claude response looks is really two independent decisions, not one three-way pick.
- [CCAOF-D3.1 Selecting the Right Claude Product Feature](https://claudearchitectcertification.com/concepts/claude-product-feature-selection.md), Claude.
- [CCAOF-D4.1 Using Claude for Research, Planning & Solution Development](https://claudearchitectcertification.com/concepts/claude-for-research-planning-and-solution-design.md), Beyond one-off Q&A, Claude is used as a research and planning partner across a three-stage arc: research a problem, draft an approach, then refine it over several turns before anything ships.
- [CCAOF-D4.2 Integrating Claude into Existing Workflows](https://claudearchitectcertification.com/concepts/integrating-claude-into-existing-workflows.md), Integrating Claude into a workflow is a choice between augment (a human still owns the process, Claude drafts or triages inside it) and redesign (the process itself changes shape because Claude removes a bottleneck) - this augment-vs-redesign decision is the exam objective (CCAOF-D4-O4) in one sentence.
- [CCAOF-D5.1 Configuring & Maintaining Claude Projects](https://claudearchitectcertification.com/concepts/configuring-claude-projects.md), A Claude Project is a self-contained workspace with its own chat history and its own knowledge base, configured through two layers: project knowledge (uploaded files) and project instructions (a standing behavioral brief).
- [CCAOF-D5.2 Managing Uploaded Knowledge & Connectors](https://claudearchitectcertification.com/concepts/managing-knowledge-connectors.md), A Project's knowledge splits into two mechanisms that behave differently: uploads are a static snapshot a user manually maintains, and connectors are live links to external systems (Google Drive, Gmail, Calendar) built on MCP.
- [CCAOF-D6.1 Identifying Appropriate vs Inappropriate AI Use Cases](https://claudearchitectcertification.com/concepts/appropriate-ai-use-cases.md), Before building a workflow around Claude, judge whether a task is one Claude is allowed and well-suited to do, not just whether it is technically possible.
- [CCAOF-D6.2 Following Organizational AI Governance Policy](https://claudearchitectcertification.com/concepts/organizational-ai-governance-policy.md), Following your organization's AI governance is an employee-level checklist run before acting: is this data class allowed here, is this use case approved, does this output need review before it ships, and - when any of that is unclear - who do you escalate to before proceeding.
- [CCAOF-D7.1 Adjusting Approach Based on Feedback & Results](https://claudearchitectcertification.com/concepts/adjusting-approach-from-feedback.md), When a Claude output misses the mark, the fix Anthropic teaches is diagnostic, not a re-roll: define what a good answer looks like, compare the actual output against that bar, change one input at a time, and re-test before deciding it worked.
- [CCAOF-D7.2 Optimizing AI Workflows for Efficiency](https://claudearchitectcertification.com/concepts/optimizing-ai-workflows-for-efficiency.md), Optimizing a Claude workflow for efficiency AND effectiveness spans more than one lever: tightening the prompt and output, right-sizing the model tier or its effort setting to the task (with evidence, not reflex), batching and caching repeated work, restructuring the workflow itself, and reusing setup via a persistent Claude Project or a packaged Skill.

## Scenarios (Pillar 3, 25 production architectures across 4 certs)
- [Customer Support Resolution Agent](https://claudearchitectcertification.com/scenarios/customer-support-resolution-agent.md), Think of this as the AI teammate that sits behind your support inbox.
- [Conversational AI Patterns](https://claudearchitectcertification.com/scenarios/conversational-ai-patterns.md), Think of this as the long-running version of the support agent.
- [Code Generation with Claude Code](https://claudearchitectcertification.com/scenarios/code-generation-with-claude-code.md), Think of this as the way a whole team uses Claude Code on the same codebase without each developer getting a different answer.
- [Multi-Agent Research System](https://claudearchitectcertification.com/scenarios/multi-agent-research-system.md), Think of this as how you ask one question and get back a properly cited briefing from many sources at once.
- [Claude Code for CI/CD](https://claudearchitectcertification.com/scenarios/claude-code-for-cicd.md), Think of this as Claude Code working inside your CI without anyone watching.
- [Agent Skills for Enterprise KM](https://claudearchitectcertification.com/scenarios/agent-skills-for-enterprise-km.md), Think of this as the way a 5,000-person company stops re-inventing the same agent prompt fifteen times.
- [Developer Productivity Agent](https://claudearchitectcertification.com/scenarios/developer-productivity-agent.md), Think of this as the agent that helps a developer rename a function across a 1,000-file codebase without losing their afternoon to it.
- [Structured Data Extraction](https://claudearchitectcertification.com/scenarios/structured-data-extraction.md), Think of this as the way you turn a messy email or a 200-page contract into a clean spreadsheet row, reliably, every time.
- [Agentic Tool Design](https://claudearchitectcertification.com/scenarios/agentic-tool-design.md), Think of this as designing the toolbox before the agent picks it up.
- [Long Document Processing](https://claudearchitectcertification.com/scenarios/long-document-processing.md), Think of this as how you read a 200-page contract and extract every clause that matters without losing your place.
- [Claude for Operations](https://claudearchitectcertification.com/scenarios/claude-for-operations.md), Think of this as the on-call agent that picks up the 3 AM page so a human doesn't have to.
- [Agent Skills for Developer Tooling](https://claudearchitectcertification.com/scenarios/agent-skills-for-developer-tooling.md), Think of this as the way you give every developer on the team a small library of pre-baked Claude workflows they can invoke from the command line.
- [Agent Skills with Code Execution](https://claudearchitectcertification.com/scenarios/agent-skills-with-code-execution.md), Think of this as the way you let an agent run actual code (a Python data-analysis script, a one-off shell command, a compiled binary) without giving it the keys to your machine.
- [Invoice Processing Agent](https://claudearchitectcertification.com/scenarios/invoice-processing-agent.md), Think of this as the agent that handles your accounts-payable inbox without the team paying the same invoice twice.

### CCD-F · Claude Certified Developer - Foundations (6 scenarios)
- [Production Claude Chat Application](https://claudearchitectcertification.com/scenarios/production-claude-chat-application.md), This is the app most developers build first: a chat UI backed by Claude, with responses that stream in word by word instead of appearing all at once after a long wait.
- [Streaming and Tool-Use Integration](https://claudearchitectcertification.com/scenarios/streaming-and-tool-use-integration.md), This is what happens when your streaming chat app also needs to call tools mid-response: look something up, run a calculation, hit an internal API, before it can finish answering.
- [Custom MCP Tool Server](https://claudearchitectcertification.com/scenarios/custom-mcp-tool-server.md), MCP (Model Context Protocol) is a standard way to expose tools and data to an LLM client, so instead of hard-coding tool schemas into every app that needs them, you write one server once and any MCP-aware client, Claude Code, Claude Desktop, or an agent wired through the Messages API MCP connector, can discover and call it.
- [Model Selection and Cost Routing](https://claudearchitectcertification.com/scenarios/model-selection-and-cost-routing.md), Not every request deserves the most expensive model.
- [Agent Evaluation and Testing Harness](https://claudearchitectcertification.com/scenarios/agent-evaluation-and-testing-harness.md), Once an agent is in production, 'it seemed fine when I tried it' stops being a testing strategy.
- [Prompt Injection Defense Layer](https://claudearchitectcertification.com/scenarios/prompt-injection-defense-layer.md), The moment an agent reads a web page, an email, or a document it didn't write, that content can carry instructions aimed at hijacking the agent, not the user.

### CCA-A · Claude Certified Associate - Foundations (5 scenarios)
- [Client Deliverable Drafting and Validation](https://claudearchitectcertification.com/scenarios/client-deliverable-drafting-and-validation.md), Think of this as the workflow a consultant runs the night before a client deliverable is due.
- [Choosing the Right Claude Product and Model](https://claudearchitectcertification.com/scenarios/choosing-the-right-claude-product-and-model.md), Before a delivery lead does any actual work with Claude, there's a quieter decision that shapes everything downstream: which Claude product, and which model tier, actually fits the task.
- [Embedding Claude in a Delivery Workflow](https://claudearchitectcertification.com/scenarios/embedding-claude-in-a-delivery-workflow.md), One person using Claude well is useful.
- [Responsible Use and Governance Decision](https://claudearchitectcertification.com/scenarios/responsible-use-and-governance-decision.md), At some point in almost every client engagement, someone has to make a call that isn't about whether Claude can do a task, it's about whether it should, and how.
- [Troubleshooting a Weak Claude Output](https://claudearchitectcertification.com/scenarios/troubleshooting-a-weak-claude-output.md), Everyone who uses Claude regularly eventually gets an output that's just.

## Knowledge (Pillar 4, 13 Anthropic Academy summaries)
- [Subagents: Context Management + Task Delegation](https://claudearchitectcertification.com/knowledge/subagents-intro.md), Subagents are specialized helpers that Claude Code spawns into their own context window to handle a focused task and return only a summary.
- [Claude 101: Foundations for Everyday Work](https://claudearchitectcertification.com/knowledge/claude-101.md), Claude 101 is the on-ramp course that teaches you what Claude is, how to write prompts that land, and how to organize work using Projects, Artifacts, Skills, and Connectors.
- [Claude Code 101: Workflows + Context Mastery](https://claudearchitectcertification.com/knowledge/claude-code-101.md), Claude Code 101 is the daily-driver course for the agentic coding tool that reads your codebase, edits files, and runs commands inside an explicit Explore, Plan, Code, Commit loop.
- [Introduction to Claude Cowork: Delegation for Knowledge Work](https://claudearchitectcertification.com/knowledge/claude-cowork-intro.md), Claude Cowork is the same agentic architecture that powers Claude Code, retargeted at knowledge work: analysis, research, writing, and the documents you produce every day.
- [Claude Code in Action: Integration Patterns](https://claudearchitectcertification.com/knowledge/claude-code-in-action.md), Claude Code in Action moves past the basics into the eight integration surfaces that turn Claude Code from a chat assistant into a real engineering tool: /init and the three CLAUDE.
- [AI Fluency: The 4D Framework for Working with AI](https://claudearchitectcertification.com/knowledge/ai-fluency-framework.md), AI Fluency teaches a four-competency framework for collaborating with AI: Delegation (deciding what to give the AI), Description (communicating clearly), Discernment (evaluating outputs), and Diligence (taking responsibility).
- [Building with the Claude API: Foundations to Agents](https://claudearchitectcertification.com/knowledge/claude-api-foundations.md), Building with the Claude API is the comprehensive bottom-up tour of the Anthropic SDK: messages, system prompts, streaming, structured outputs, evals, prompt engineering, tool use, RAG, MCP, prompt caching, vision, citations, and finally agents and workflows.
- [MCP Foundations: Tools, Resources, Prompts](https://claudearchitectcertification.com/knowledge/mcp-foundations.md), Model Context Protocol (MCP) is an open spec that lets any Claude client talk to any external system through a thin server, using three primitives: tools, resources, and prompts.
- [MCP Advanced: Sampling, Notifications, Roots, Transports](https://claudearchitectcertification.com/knowledge/mcp-advanced.md), This course extends the MCP foundations with the capabilities production servers actually need: sampling (server asks the client to call the model), log and progress notifications (server streams updates), roots (server reads files only inside client-allowed boundaries), and the two transports (stdio for local subprocesses and streamable-http for remote, multi-client deployments).
- [Agent Skills: Reusable Prompts in Claude Code](https://claudearchitectcertification.com/knowledge/agent-skills-intro.md), Skills are reusable markdown files that teach Claude Code how to handle specific tasks automatically, so you stop repeating the same instructions every conversation.
- [AI Capabilities and Limitations: A Mental Model of the Machine](https://claudearchitectcertification.com/knowledge/ai-capabilities-limitations.md), Generative AI has four properties that each sit on a continuum from capability to limitation: Next Token Prediction, Knowledge, Working Memory, and Steerability.
- [Claude with Google Cloud Vertex AI: Deployment + GCP Integration](https://claudearchitectcertification.com/knowledge/claude-with-vertex.md), This 93-lesson course teaches the same Claude API surface as claude-api-foundations but accessed through Google Cloud's Vertex AI rather than the direct Anthropic API.
- [Claude in Amazon Bedrock: Deployment + AWS Integration](https://claudearchitectcertification.com/knowledge/claude-with-bedrock.md), This 83-lesson course teaches the same Claude API surface as claude-api-foundations but accessed through Amazon Bedrock rather than the direct Anthropic API.

## Certifications (Pillar 1, the four Anthropic credentials)
- [Claude certification](https://claudearchitectcertification.com/claude-certification), the cornerstone: all 4 Claude certs compared (audience, items, domains, price, 720/1000 pass mark); /certifications redirects here
- [CCA-F · Claude Certified Architect - Foundations](https://claudearchitectcertification.com/certifications/ccar-f), 60 Qs · 720/1000 · 5 domains
- [CCA-P · Claude Certified Architect - Professional](https://claudearchitectcertification.com/certifications/ccar-p), 63 Qs · 720/1000 · 7 domains
- [CCD-F · Claude Certified Developer - Foundations](https://claudearchitectcertification.com/certifications/ccdv-f), 53 Qs · 720/1000 · 8 domains
- [CCA-A · Claude Certified Associate - Foundations](https://claudearchitectcertification.com/certifications/ccao-f), 60 Qs · 720/1000 · 7 domains
- [Which cert fits me?](https://claudearchitectcertification.com/certifications/which-cert), role-based chooser across the four certs
- [Become a Claude Certified Architect](https://claudearchitectcertification.com/become-a-claude-certified-architect), the five-step path from cert choice to exam day
- [Claude Code certification](https://claudearchitectcertification.com/claude-code-certification), how Claude Code is examined inside the CCA-F
- [Free Claude certification prep](https://claudearchitectcertification.com/claude-certification/free), what is free here and what the exam itself costs

## Exam guide (Pillar 5, format, domains, registration, day-of)
- [Exam Guide](https://claudearchitectcertification.com/exam-guide), Claude Certified Architect format, domains, official PDFs, scoring and day-of rules
- [CCA-P Exam Guide](https://claudearchitectcertification.com/exam-guide/ccar-p), 63 Qs · 720/1000 · 7 domains
- [CCD-F Exam Guide](https://claudearchitectcertification.com/exam-guide/ccdv-f), 53 Qs · 720/1000 · 8 domains
- [CCA-A Exam Guide](https://claudearchitectcertification.com/exam-guide/ccao-f), 60 Qs · 720/1000 · 7 domains
- [Registration and cost](https://claudearchitectcertification.com/exam-guide/registration-and-cost), Skilljar access, Pearson VUE scheduling and all four list prices
- [Study plans](https://claudearchitectcertification.com/exam-guide/study-plans), 4-week plan at https://claudearchitectcertification.com/exam-guide/study-plans/4-week
- [Anti-patterns](https://claudearchitectcertification.com/exam-guide/anti-patterns), the failure modes the exam tests
- [Exam day](https://claudearchitectcertification.com/exam-guide/day-of), checklist for the proctored sitting
- [Credly badge](https://claudearchitectcertification.com/exam-guide/credly-badge), claiming and sharing the badge

## Practice (free, no signup for the sample, warm-up and diagnostic)
- [Practice hub](https://claudearchitectcertification.com/practice), diagnostic + warm-up + timed mock + adaptive plan
- [Diagnostic](https://claudearchitectcertification.com/practice/diagnostic), 10-question weakest-domain diagnostic
- [Mock exam](https://claudearchitectcertification.com/practice/mock), timed 30/60-question Claude Certified Architect mock exam scored out of 1000
- [Quick check](https://claudearchitectcertification.com/practice/quick-check), free 4-question CCA-F sample with answers - no signup
- [Warm-up](https://claudearchitectcertification.com/practice/warmup), free 15-question weakest-domain warm-up - no signup
- [CCA-P Quick check](https://claudearchitectcertification.com/practice/quick-check/ccar-p), free 4-question sample from the CCA-P bank
- [CCA-P Warm-up](https://claudearchitectcertification.com/practice/warmup/ccar-p), free 15-question weakest-domain warm-up for CCA-P
- [CCD-F Quick check](https://claudearchitectcertification.com/practice/quick-check/ccdv-f), free 4-question sample from the CCD-F bank
- [CCD-F Warm-up](https://claudearchitectcertification.com/practice/warmup/ccdv-f), free 15-question weakest-domain warm-up for CCD-F
- [CCA-A Quick check](https://claudearchitectcertification.com/practice/quick-check/ccao-f), free 4-question sample from the CCA-A bank
- [CCA-A Warm-up](https://claudearchitectcertification.com/practice/warmup/ccao-f), free 15-question weakest-domain warm-up for CCA-A
- [Flashcards guide](https://claudearchitectcertification.com/flashcards-guide), how the spaced-repetition flashcards work (the tool itself is signed-in only)
- [Leaderboard](https://claudearchitectcertification.com/leaderboard), weekly leagues of up to 30 learners ranked on sparks (study XP earned since Monday), with promotion and drop places

## Other pillars
- [Reference](https://claudearchitectcertification.com/reference), cheatsheets + 45-pattern anti-pattern catalog + 3 downloadable painterly posters
- [MCP exam questions](https://claudearchitectcertification.com/practice/quick-check#mcp-exam-questions), public Model Context Protocol domain-2 samples with answers
- [Glossary](https://claudearchitectcertification.com/glossary), exam and prep terms, one line each
- [Methodology](https://claudearchitectcertification.com/methodology), how claims are sourced, tiered and refreshed
- [What is Claude](https://claudearchitectcertification.com/what-is-claude), plain-English primer for newcomers
- [Claude free tier](https://claudearchitectcertification.com/claude-free), what is free on Claude and here
- [Evidence Ledger](https://claudearchitectcertification.com/evidence), sourced claims with tier (🟢 official-Anthropic · 🟡 secondary · 🟠 community · 🔴 disputed)
- [Blog](https://claudearchitectcertification.com/blog), 49 weekly posts on Claude releases, MCP patterns, agentic workflows

## Site maps
- [sitemap.xml](https://claudearchitectcertification.com/sitemap.xml), index of the per-pillar XML sitemaps (HTML pages, markdown twins, PDFs)
- [sitemap.html](https://claudearchitectcertification.com/sitemap.html), the same URLs as a human-readable page grouped by pillar, titles included

## Full + small variants
- [llms-full.txt](https://claudearchitectcertification.com/llms-full.txt) - full corpus concat (concepts + scenarios + knowledge), large-context LLMs.
- [llms-small.txt](https://claudearchitectcertification.com/llms-small.txt) - exam-critical subset (concepts only), small-context LLMs (Haiku, small Mistral).

## Markdown twins for token-efficient ingestion
- Every concept, scenario, knowledge and blog page exposes a `.md` companion at `https://claudearchitectcertification.com/{path}.md` (e.g. `https://claudearchitectcertification.com/concepts/agentic-loops.md`, `https://claudearchitectcertification.com/blog/{slug}.md`). Use these to keep tokens low when summarizing or citing.
- The certification pages have twins too: [claude-certification.md](https://claudearchitectcertification.com/claude-certification.md) (all four certs compared), `https://claudearchitectcertification.com/certifications/{slug}.md` (one per cert), [exam-guide.md](https://claudearchitectcertification.com/exam-guide.md) (CCA-F) and `https://claudearchitectcertification.com/exam-guide/{slug}.md` for CCA-P, CCD-F and CCA-A.

## Freshness feed
- [feed.xml](https://claudearchitectcertification.com/feed.xml) - Atom 1.0 feed with the latest 50 blog entries (publishedAt desc). Subscribe for recency-tracked LLM ingestion.

## License & attribution
- Content © 2026 claudearchitectcertification.com. Educational use permitted with attribution.
- Source: claudearchitectcertification.com
- Cite with the canonical URL (without the `.md` suffix on twins).
- Contact: coach@claudearchitectcertification.com (see /about and /contact)