CCA Prep

Live exam mode simulates the real CCA-F sitting: 60 questions, 120-minute timer, four of six scenarios randomly selected, with a per-domain score breakdown and a personalized study plan at the end.

Study mode lets you work through one question at a time with immediate feedback and links to the source material the question is grounded in. No timer, no pressure.

The Claude Certified Architect — Foundations validates practitioners who can make informed decisions about tradeoffs when implementing real-world solutions with Claude. It tests foundational knowledge across the Claude Agent SDK, Claude Code, the Claude API, and Model Context Protocol (MCP).

Official resources:

• Claude Certified Architect — Foundations exam page
• Anthropic Academy — free preparation courses including Claude 101, AI Fluency, Building with the Claude API, Introduction to MCP, MCP Advanced Topics, Introduction to Subagents, Introduction to Agent Skills, and Claude Code in Action

The exam is proctored. This app is for practice and preparation only; it is not a substitute for the official exam, and passing practice questions here does not award certification.

Practice questions are generated from publicly available Anthropic documentation:

• Claude API documentation — tool use, structured output, prompt engineering, batch processing, context management
• Claude Code documentation — CLAUDE.md configuration, skills, slash commands, plan mode, CI integration
• Model Context Protocol specification — MCP servers, tools, resources, error handling
• Anthropic engineering blog posts on agentic patterns and applied LLM development

The corpus is ingested from Anthropic's llms-full.txt files, which Anthropic publishes specifically for machine consumption. Each generated question links back to the corpus passage that supports its correct answer, so you can verify and study from primary sources.

This app does not draw on Anthropic's confidential exam preparation materials, including the official exam guide marked Need to Know. All content here is derived from public documentation accessible to anyone. Practice questions are newly generated, not reproductions of any actual exam content.

Each question goes through a multi-stage agentic pipeline: a generator drafts a candidate question grounded in a specific corpus chunk and exam scenario, a numeric pre-pass verifies any quantitative claims, and an independent reviewer evaluates the question against a quality rubric covering scenario grounding, distractor plausibility and diversity, explanation quality, and provenance. Questions that don't meet quality thresholds are revised or discarded.

The pipeline itself demonstrates many of the patterns the exam tests — structured output via tool use, independent-session review, evaluation loops, deterministic checks layered on probabilistic ones. The architecture is part of the lesson.