Ten CCA Foundations-Style Practice Questions, With Explanations
Ten original scenario questions across tool authorization, orchestration, Claude Code enforcement, structured output and context strategy — each with the reasoning behind the best answer.
Before you start
Anthropic publicly describes Claude Certified Architect, Foundations as its first technical certification and as an exam for solution architects building production applications with Claude. Anthropic's public announcement does not disclose the live exam's question count, duration, passing score, price, domain weights, delivery provider, or exact question style. Confirm those details in the current Partner Academy, candidate guide, registration portal, and appointment email.
The questions below are original study questions. They are not copied, recalled, or reconstructed from the live exam. They test production judgment using current Anthropic documentation.
| Rule | What to do | Why it matters |
|---|---|---|
| Commit first | Pick one option before you reveal the explanation. | Reading the rationale first turns a diagnostic into passive reading. |
| Record confidence | Note whether each answer was certain or a guess. | A confident wrong answer is a misconception; a lucky guess is still a gap. |
| Explain the rejects | Say out loud why each wrong option fails. | The live exam separates candidates on plausible distractors, not obvious ones. |
| Convert to practice | Turn every miss into one small implementation. | Architecture judgment is built by shipping, not by re-reading. |
What this set covers
| # | Topic | Judgment being tested |
|---|---|---|
| 1 | Consequential tool actions | Where an authorization boundary belongs |
| 2 | Parallel vs sequential work | Dependency-aware orchestration |
| 3 | CLAUDE.md vs enforcement | Context guidance vs hard security controls |
| 4 | Skills vs hooks | Invoked workflow vs automatic lifecycle event |
| 5 | Machine-consumable output | Schema conformance plus semantic validation |
| 6 | Missing source information | Honest representation of absent data |
| 7 | Tool authorization | Untrusted model-generated arguments |
| 8 | MCP capability type | Resources vs tools vs prompts |
| 9 | Rate limit handling | Retry-After, bounded backoff, deadlines |
| 10 | Prompt caching | Stable prefix before dynamic content |
Reading your score
| Score | What it signals | Next action |
|---|---|---|
| 9-10 | Strong on these specific scenarios. | Verify you can explain every rejected option, then move to full-length sets. |
| 7-8 | Principles mostly held, one or two gaps. | Review the missed principle and build one small implementation. |
| 5-6 | Recognition without reliable judgment. | Revisit the official documentation below before attempting larger mock sets. |
| 0-4 | Foundations not yet in place. | Build hands-on experience with tools, permissions and structured outputs, then repeat with new questions. |
Error log template
Copy this table once per missed question. The Principle column is the part that transfers to the exam.
| Field | Your entry |
|---|---|
| Question | |
| My answer | |
| Correct answer | |
| Why I chose incorrectly | |
| Principle to remember | |
| Official page reviewed | |
| One implementation exercise |
Official sources
- Certification announcement: https://www.anthropic.com/news/claude-partner-network
- Claude Code security: https://docs.anthropic.com/en/docs/claude-code/security
- Claude Code memory and CLAUDE.md: https://docs.anthropic.com/en/docs/claude-code/memory
- Claude Code skills: https://docs.anthropic.com/en/docs/claude-code/skills
- Structured outputs and platform features: https://docs.anthropic.com/en/docs/build-with-claude/overview
- Tool use: https://docs.anthropic.com/en/docs/build-with-claude/tool-use/overview
- MCP: https://docs.anthropic.com/en/docs/agents-and-tools/mcp
- Rate limits: https://docs.anthropic.com/en/api/rate-limits
- Prompt caching: https://docs.anthropic.com/en/docs/build-with-claude/prompt-caching