# Claude Cert Prep

> Free, independent prep for the Claude Certified certifications (Architect, Developer, and Associate). In-depth study guides, mock exams, and domain-weighted practice with per-objective tracking — starting with the Architect – Foundations (CCA-F / CCAR-F) track.

Independent and unofficial — not affiliated with, endorsed by, or authorized by Anthropic. All questions and explanations are original; no real exam items are reproduced. Every concept page below has a raw-markdown twin at its `/md` URL for clean ingestion.

## Claude Certified Architect – Foundations (CCA-F)

- [Certification overview](https://claude.ajithkumarr.com/certifications/cca-f)
- [Study guides](https://claude.ajithkumarr.com/certifications/cca-f/guides)
- [Mock exams](https://claude.ajithkumarr.com/certifications/cca-f/mock-exam)

### CCA-F guides
- [The Complete CCA-F Study Guide: Passing the Claude Certified Architect – Foundations Exam](https://claude.ajithkumarr.com/certifications/cca-f/guides/cca-f-complete-study-guide): What the CCA-F exam actually tests, how it's weighted across five domains and 29 objectives, and a realistic plan to prepare for it.
- [How the CCA-F Exam Thinks: The Architect's Mindset, Traps, and Decision Framework](https://claude.ajithkumarr.com/certifications/cca-f/guides/how-the-cca-f-exam-thinks): The CCA-F is a scenario-based architecture-judgment exam, not an API memory test. Learn the one mental model, the elimination framework, and the five traps that decide most questions.
- [CCA-F Domain 1 Study Guide: Agentic Architecture & Orchestration](https://claude.ajithkumarr.com/certifications/cca-f/guides/agentic-architecture-and-orchestration): The heaviest domain on the Claude Certified Architect – Foundations exam (27%). Master the agentic loop, stop_reason handling, structured handoffs, dynamic decomposition, and deterministic hook enforcement.
- [CCA-F Domain 2 Study Guide: Tool Design & MCP Integration](https://claude.ajithkumarr.com/certifications/cca-f/guides/tool-design-and-mcp-integration): How to design tool schemas, error contracts, and MCP server configurations that let Claude choose, call, and recover from tools reliably.
- [CCA-F Domain 3 Study Guide: Claude Code Configuration & Workflows](https://claude.ajithkumarr.com/certifications/cca-f/guides/claude-code-configuration-and-workflows): Configure Claude Code the way it actually behaves — knowing when guidance is a suggestion the model may follow and when it is a rule the harness enforces.
- [CCA-F Domain 4 Study Guide: Prompt Engineering & Structured Output](https://claude.ajithkumarr.com/certifications/cca-f/guides/prompt-engineering-and-structured-output): Choosing and enforcing reliable structured output on the Claude Messages API — from tool schemas and tool_choice to extraction design, feedback loops, and picking synchronous vs. batch processing.
- [CCA-F Domain 5 Study Guide: Context Management & Reliability](https://claude.ajithkumarr.com/certifications/cca-f/guides/context-management-and-reliability): Sustaining coherent agents across context limits, stateless APIs, escalation calls, and risk-based human review.
- [Frequently Confused Concepts — Agentic Architecture & Orchestration (CCA-F Domain 1)](https://claude.ajithkumarr.com/certifications/cca-f/guides/confused-concepts-agentic-orchestration): Five look-alike pairs from Domain 1, compared side by side — so you pick the architecturally-correct orchestration mechanism for a scenario instead of pattern-matching on vocabulary.
- [Frequently Confused Concepts — Tool Design & MCP Integration (CCA-F Domain 2)](https://claude.ajithkumarr.com/certifications/cca-f/guides/confused-concepts-tools-mcp): Five look-alike mechanisms in Anthropic tool use and MCP — and how to pick the right one for a scenario instead of memorizing definitions.
- [Frequently Confused Concepts — Claude Code Configuration (CCA-F Domain 3)](https://claude.ajithkumarr.com/certifications/cca-f/guides/confused-concepts-claude-code): CLAUDE.md, Rules, Skills, slash commands, hooks, permissions, and the planning modes all shape how Claude Code behaves — but they sit at different layers. Learn the mental model so any scenario instantly maps to the right mechanism.
- [Frequently Confused Concepts — Prompt Engineering & Structured Output (CCA-F Domain 4)](https://claude.ajithkumarr.com/certifications/cca-f/guides/confused-concepts-prompt-output): Six prompting and output techniques that look interchangeable on paper but answer different questions in practice — pick the right one for the scenario, not the one you saw last.
- [Frequently Confused Concepts — Context Management & Reliability (CCA-F Domain 5)](https://claude.ajithkumarr.com/certifications/cca-f/guides/confused-concepts-context-reliability): Six session-management and reliability distinctions that decide the right answer the moment you spot the scenario — resume vs compact vs new, forking vs resuming, transcript vs summary, window vs memory, window vs session, and retry vs escalation.
- [CCA-F Production Patterns & Deep Cuts: The Specifics the Exam Loves to Test](https://claude.ajithkumarr.com/certifications/cca-f/guides/cca-f-production-patterns-deep-cuts): The concrete, easily-missed patterns behind the scenarios — PostToolUse normalization, Case-Facts persistence, context: fork, headless CI flags, the Interview pattern, and the myths to unlearn.
- [CCA-F Glossary: Key Terms for the Claude Certified Architect – Foundations Exam](https://claude.ajithkumarr.com/certifications/cca-f/guides/cca-f-glossary-key-terms): A fast reference to the vocabulary the CCA-F scenarios assume you know — the agentic loop, MCP, hooks, tool_choice, structured output, the stateless API, and more.

## Claude Certified Developer – Foundations (CCDV-F)

- [Certification overview](https://claude.ajithkumarr.com/certifications/ccdv-f)
- [Study guides](https://claude.ajithkumarr.com/certifications/ccdv-f/guides)
- [Mock exams](https://claude.ajithkumarr.com/certifications/ccdv-f/mock-exam)

### CCDV-F guides
- [The Complete CCDV-F Study Guide: Claude Certified Developer – Foundations](https://claude.ajithkumarr.com/certifications/ccdv-f/guides/ccdv-f-complete-study-guide): What the Developer exam actually tests, how its eight domains are weighted, and a study plan built around the heaviest one: shipping real applications against the Claude API.
- [CCDV-F Domain 1: Agents and Workflows](https://claude.ajithkumarr.com/certifications/ccdv-f/guides/ccdv-f-agents-and-workflows): Why the exam cares more about when you reach for an agent than how you wire one up — the workflow-vs-agent distinction, the five named workflow patterns, and where the Claude Agent SDK fits next to raw API tool-use and Managed Agents.
- [CCDV-F Domain 2: Applications and Integration](https://claude.ajithkumarr.com/certifications/ccdv-f/guides/ccdv-f-applications-and-integration): The single heaviest domain on the exam: how to scope, build, and ship a production Claude integration — request mechanics, streaming vs. batch, the retryable-vs-permanent error trap, and the engineering discipline around it.
- [CCDV-F Domain 5: Model Selection and Optimization](https://claude.ajithkumarr.com/certifications/ccdv-f/guides/ccdv-f-model-selection-and-optimization): How to reason about tokens, context windows, and adaptive thinking like a developer — then turn that into a defensible model choice among Opus, Sonnet, Haiku, and Fable, and a cost model that survives contact with a real workload.
- [CCDV-F Domain 6: Prompt and Context Engineering](https://claude.ajithkumarr.com/certifications/ccdv-f/guides/ccdv-f-prompt-and-context-engineering): Context is a budget, not a bottomless buffer. How to manage it deliberately, write instructions Claude will actually follow, and never trust model output without validating it — the three sub-skills behind Domain 6's 11% of the CCDV-F exam.
- [CCDV-F Domain 7: Security and Safety](https://claude.ajithkumarr.com/certifications/ccdv-f/guides/ccdv-f-security-and-safety): Prompt injection vs. jailbreak, the tool_result-only rule for untrusted content, least-privilege sandboxing, Claude Hooks as deterministic enforcement, and the three ways to authenticate to the Claude API.
- [CCDV-F Domain 8: Tools and MCPs](https://claude.ajithkumarr.com/certifications/ccdv-f/guides/ccdv-f-tools-and-mcps): How to define tools Claude actually calls, run the tool_use/tool_result loop correctly, and choose between built-in tools, custom tools, Skills, and MCP servers — including what changed in the MCP spec's July 2026 revision.
- [CCDV-F Domain 3: Claude Code](https://claude.ajithkumarr.com/certifications/ccdv-f/guides/ccdv-f-claude-code): How CLAUDE.md, settings.json, permission modes, hooks, Skills, subagents, and headless mode fit together to run Claude Code safely on a real project — and the one setting that changed its default behavior in mid-2026.
- [CCDV-F Domain 4: Eval, Testing, and Debugging](https://claude.ajithkumarr.com/certifications/ccdv-f/guides/ccdv-f-eval-testing-and-debugging): The exam's smallest domain tests one skill: correctly isolating which layer actually failed — API, tool, prompt, or model — from a single observed symptom, plus the deterministic-vs-model-based evaluation and regression-testing discipline that keeps Claude apps stable as they change.
- [Top CCDV-F Mistakes (and the Traps That Cost Real Points)](https://claude.ajithkumarr.com/certifications/ccdv-f/guides/ccdv-f-top-mistakes): An independent, unofficial field guide to the implementation traps that catch even experienced developers on the Claude Certified Developer – Foundations exam.
- [Frequently Confused Concepts — Claude Certified Developer Foundations (CCDV-F)](https://claude.ajithkumarr.com/certifications/ccdv-f/guides/ccdv-f-confused-concepts): Six implementation choices where several options genuinely work — the mental model that decides which one the exam rewards.

## Claude Certified Architect – Professional (CCAR-P)

- [Certification overview](https://claude.ajithkumarr.com/certifications/ccar-p)
- [Study guides](https://claude.ajithkumarr.com/certifications/ccar-p/guides)
- [Concepts glossary](https://claude.ajithkumarr.com/certifications/ccar-p/concepts)
- [Mock exams](https://claude.ajithkumarr.com/certifications/ccar-p/mock-exam)
- [Enterprise case studies](https://claude.ajithkumarr.com/certifications/ccar-p/case-studies)
- [Practice by domain](https://claude.ajithkumarr.com/certifications/ccar-p/practice)

### CCAR-P guides
- [The Complete CCAR-P Study Guide: Claude Certified Architect – Professional](https://claude.ajithkumarr.com/certifications/ccar-p/guides/ccar-p-complete-study-guide): What the Professional architect exam tests, how its seven domains are weighted, and a plan to prepare for the advanced tier above Foundations.
- [How the CCAR-P Exam Thinks: A Judgment Exam, Not a Recall Exam](https://claude.ajithkumarr.com/certifications/ccar-p/guides/how-the-ccar-p-exam-thinks): The Professional architect exam rewards trade-off reasoning and root-cause judgment. Here's the mental model that separates a pass — and the traps built into every question.
- [CCAR-P Domain 1: Solution Design & Architecture](https://claude.ajithkumarr.com/certifications/ccar-p/guides/ccar-p-solution-design-architecture): How to make architect-level design decisions for Claude systems at enterprise scale: agentic vs. deterministic, model-tier selection, terminal guarantees, and designing for failure.
- [CCAR-P Domain 2: Claude Models, Prompting & Context Engineering](https://claude.ajithkumarr.com/certifications/ccar-p/guides/ccar-p-models-prompting-context): An enterprise-architect's guide to choosing and routing Claude models, engineering prompt caching and long-context strategy, and building structured-output pipelines that stay reliable at scale.
- [CCAR-P Domain 3: Integration — RAG, MCP, Auth & Observability](https://claude.ajithkumarr.com/certifications/ccar-p/guides/ccar-p-integration-rag-mcp): The heaviest domain on the exam. An enterprise architect's job is not to make one call succeed — it is to design retrieval, tool surfaces, identity, and telemetry that stay correct, cheap, and debuggable under thousands of concurrent agent runs.
- [CCAR-P Domain 4: Evaluation, Testing & Optimization](https://claude.ajithkumarr.com/certifications/ccar-p/guides/ccar-p-evaluation-testing-optimization): How an enterprise architect designs evals, tests prompts and agents like software, and drives down cost and latency without eroding quality — the 16% of the CCAR-P blueprint that separates demos from production.
- [CCAR-P Domain 5: Governance, Safety & Risk Management](https://claude.ajithkumarr.com/certifications/ccar-p/guides/ccar-p-governance-safety-risk): Enforcing hard rules deterministically, defending against prompt injection, and building auditable, least-privilege agent systems for the Claude Certified Architect - Professional exam.
- [CCAR-P Domain 6: Stakeholder Communication & Lifecycle Management](https://claude.ajithkumarr.com/certifications/ccar-p/guides/ccar-p-stakeholder-lifecycle): How enterprise architects elicit requirements, communicate probabilistic-system tradeoffs to non-technical stakeholders, and steward Claude-powered features from POC through production to deprecation.
- [CCAR-P Domain 7: Developer Productivity & Operational Enablement](https://claude.ajithkumarr.com/certifications/ccar-p/guides/ccar-p-developer-productivity-ops): How an enterprise architect scales Claude Code across a team: shared configuration, deterministic guardrails, headless CI automation, MCP distribution, and reproducible golden-path onboarding.
- [Frequently Confused Concepts — Claude Certified Architect Professional (CCAR-P)](https://claude.ajithkumarr.com/certifications/ccar-p/guides/ccar-p-confused-concepts): Six enterprise-level architecture choices where several answers look right — learn the mental model that decides which one the exam rewards.
- [Top CCAR-P Mistakes (and the Decisions That Separate a Pass)](https://claude.ajithkumarr.com/certifications/ccar-p/guides/ccar-p-top-mistakes): An independent, unofficial field guide to the judgment traps that sink otherwise-strong candidates on the Claude Certified Architect - Professional exam.
- [Judgment Checkpoints: Reasoning Through CCAR-P's Hardest Question Types](https://claude.ajithkumarr.com/certifications/ccar-p/guides/ccar-p-judgment-checkpoints): An independent, unofficial guide to the recurring decision formats on the CCAR-P exam — and the governing principles that turn each one from a guess into a defensible call.

### CCAR-P concepts (with markdown twins)
- [Agentic Loop](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/agentic-loop) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/agentic-loop/md): The repeating model-tool-observe cycle that lets Claude act, read results, and decide the next step until a goal is met.
- [Workflow vs Agent](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/workflow-vs-agent) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/workflow-vs-agent/md): Workflows follow predefined code paths; agents let the model dynamically direct its own steps and tool use.
- [Orchestrator-Worker Topology](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/orchestrator-worker-topology) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/orchestrator-worker-topology/md): A coordinator agent decomposes a task, delegates subtasks to worker subagents, and synthesizes their results into a final answer.
- [Single-Agent vs Multi-Agent](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/single-vs-multi-agent) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/single-vs-multi-agent/md): Whether one agent handles a task end-to-end or several specialized agents coordinate — a tradeoff of simplicity against parallelism and isolation.
- [Human-in-the-Loop Gate](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/human-in-the-loop-gate) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/human-in-the-loop-gate/md): A required human approval checkpoint inserted before an agent takes a consequential or irreversible action.
- [Idempotency & Retries](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/idempotency-and-retries) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/idempotency-and-retries/md): Designing agent actions so that safely retrying a failed or duplicated call produces the same effect as running it once.
- [Batch vs Synchronous Processing](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/batch-vs-synchronous) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/batch-vs-synchronous/md): Choosing between interactive real-time requests and the Batch API's cheaper, high-latency asynchronous processing for large non-urgent jobs.
- [Model selection (Opus vs Sonnet vs Haiku)](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/model-selection-opus-sonnet-haiku) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/model-selection-opus-sonnet-haiku/md): Match the task to the tier: Opus for hardest reasoning, Sonnet for balanced work, Haiku for fast, cheap, high-volume tasks.
- [No customer fine-tuning of Claude](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/no-customer-fine-tuning) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/no-customer-fine-tuning/md): You cannot retrain Claude's weights on your data; you adapt behavior through prompts, context, tools, and RAG.
- [Prompt caching](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/prompt-caching) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/prompt-caching/md): Reuse a stable, up-front prompt prefix across calls so repeated tokens are cheaper and faster to process.
- [Extended thinking](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/extended-thinking) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/extended-thinking/md): An optional mode where Claude produces internal reasoning before its answer, improving hard tasks at extra token cost.
- [Context window management](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/context-window-management) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/context-window-management/md): Actively curating what goes into the model's finite token window so the relevant signal isn't crowded out.
- [Structured output](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/structured-output) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/structured-output/md): Getting the model to return machine-parseable data (usually JSON) that conforms to a schema you define.
- [Context rot](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/context-rot) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/context-rot/md): Quality decay that sets in when a context window fills with stale, irrelevant, or contradictory tokens.
- [Model Context Protocol (MCP)](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/mcp) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/mcp/md): An open protocol that standardizes how apps expose tools, data, and prompts to LLMs over a client/server link.
- [RAG Pipeline](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/rag-pipeline) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/rag-pipeline/md): Retrieve relevant documents at query time and ground the model's answer in them instead of relying on parametric memory.
- [Chunking Strategy](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/chunking-strategy) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/chunking-strategy/md): How you split source documents into retrievable units — it sets the ceiling on everything retrieval can return.
- [Reranking](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/reranking) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/reranking/md): A precision second pass that reorders an over-fetched candidate set so the most relevant chunks land in the prompt.
- [Hybrid Retrieval](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/hybrid-retrieval) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/hybrid-retrieval/md): Combines semantic (dense vector) and keyword (sparse/BM25) search so you catch both meaning and exact terms.
- [Data-Layer ACLs](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/data-layer-acls) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/data-layer-acls/md): Enforce who can see which documents in the retrieval/data layer — never by asking the model in the prompt to filter.
- [MCP vs Direct API Integration](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/mcp-vs-direct-api) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/mcp-vs-direct-api/md): MCP is a reusable standard layer for exposing tools/data to any MCP host; a direct API call is bespoke, app-specific glue.
- [Observability & Tracing](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/observability-tracing) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/observability-tracing/md): Instrument each step of an agent/RAG run — prompts, retrievals, tool calls, tokens, latency — so you can debug and evaluate it.
- [Eval-Driven Development](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/eval-driven-development) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/eval-driven-development/md): Define measurable success criteria and an eval harness before building, then let scores gate every change.
- [LLM-as-Judge](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/llm-as-judge) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/llm-as-judge/md): Use a strong LLM to score or compare outputs against a rubric when exact-match grading is impossible.
- [Golden Dataset](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/golden-dataset) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/golden-dataset/md): A curated, versioned set of representative inputs with trusted expected outputs used to score every change.
- [Offline vs. Online Evaluation](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/offline-vs-online-eval) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/offline-vs-online-eval/md): Offline eval scores a fixed dataset before shipping; online eval measures real behavior on live traffic.
- [Retrieval Quality vs. Faithfulness](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/retrieval-vs-faithfulness) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/retrieval-vs-faithfulness/md): Separate whether RAG fetched the right context (retrieval) from whether the answer stayed true to it (faithfulness).
- [Regression Testing](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/regression-testing) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/regression-testing/md): Re-run a fixed eval suite on every change so quality drops are caught before they reach users.
- [Cost & Latency Optimization](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/cost-latency-optimization) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/cost-latency-optimization/md): Cut spend and response time without dropping below your quality bar, using evals to guard the tradeoff.
- [Escalate on Policy, Not Confidence](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/escalate-on-policy-not-confidence) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/escalate-on-policy-not-confidence/md): Route to a human on fixed policy rules, never on the model's self-reported confidence score.
- [Fail-Closed vs Fail-Open](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/fail-closed-vs-fail-open) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/fail-closed-vs-fail-open/md): When a safety check errors or is unavailable, deny the action (fail-closed) rather than allow it (fail-open).
- [Guardrail Placement](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/guardrail-placement) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/guardrail-placement/md): Guardrails belong at input screening, output screening, and tool-call authorization — a single output filter is not enough.
- [Least Privilege](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/least-privilege) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/least-privilege/md): Grant an agent, tool, or credential only the minimum access needed for its task — nothing more.
- [Prompt Injection Defense](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/prompt-injection-defense) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/prompt-injection-defense/md): Treat all retrieved and tool-returned content as untrusted data, and enforce must-hold rules in code, not the prompt.
- [Compliance Control Register](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/compliance-control-register) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/compliance-control-register/md): A living map of each obligation to a named control, an accountable owner, and the evidence that proves it works.
- [Alignment Boundary](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/alignment-boundary) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/alignment-boundary/md): The line between what model training covers and what your application layer must enforce itself.
- [PII Redaction](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/pii-redaction) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/pii-redaction/md): Detect and remove personal data before it reaches the model, logs, or downstream systems that don't need it.
- [Signal Triage](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/signal-triage) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/signal-triage/md): Deciding which incoming signals warrant escalation versus routine handling, using pre-agreed rules rather than in-the-moment judgment.
- [Reversal Cost](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/reversal-cost) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/reversal-cost/md): What it would take to undo a decision — the element most often missing from a Gains / Gives-up / Reversal-cost analysis.
- [Outcome Document](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/outcome-document) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/outcome-document/md): The auditable record of what a project achieved, anchored to a control-in-place and a named measurement owner.
- [PoC vs Pilot vs Production](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/poc-vs-pilot-vs-production) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/poc-vs-pilot-vs-production/md): Three lifecycle stages with rising evidence bars: PoC proves it can work, pilot proves it works for real users, production owns it.
- [Canary vs Blue-Green](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/canary-vs-blue-green) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/canary-vs-blue-green/md): Two rollout patterns: canary exposes a small slice first; blue-green flips all traffic between two full environments.
- [Migration Parity Eval](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/migration-parity-eval) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/migration-parity-eval/md): Proving the new system produces equivalent results to the old one before you cut over — not after.
- [Stakeholder Sign-off](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/stakeholder-sign-off) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/stakeholder-sign-off/md): The accountable owner's recorded agreement to proceed or accept — and the point where a design only the architect understands fails.
- [Skills Distribution](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/skills-distribution) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/skills-distribution/md): Publishing, versioning, and governing reusable Claude Skills across a team so upgrades and spend stay controlled.
- [Claude Code in CI/CD](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/claude-code-in-cicd) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/claude-code-in-cicd/md): Running Claude Code non-interactively in a build pipeline with guardrails on any code it changes.
- [AI Code Verification Checklist](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/ai-code-verification-checklist) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/ai-code-verification-checklist/md): A structured review pass that confirms AI-written code is actually correct, not merely syntactically valid.
- [Symptom-to-Architecture-Cause Mapping](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/symptom-to-architecture-cause) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/symptom-to-architecture-cause/md): Diagnosing an agent's production symptom by mapping it to the architectural layer that actually caused it.
- [Golden Paths](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/golden-paths) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/golden-paths/md): Opinionated, well-supported default workflows that make the safe, productive way the easy way for developers.
- [Agent SDK vs API vs Claude Code](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/agent-sdk-vs-api-vs-claude-code) · [.md](https://claude.ajithkumarr.com/certifications/ccar-p/concepts/agent-sdk-vs-api-vs-claude-code/md): Choosing the right Claude build surface: raw API, the Agent SDK harness, or the full Claude Code product.

---
Full sitemap: https://claude.ajithkumarr.com/sitemap.xml