CCA-F · Claude Certified Architect – Foundations

Cheat sheet

720 / 1000
Passing score (100 to 1000 scale)
4 / 6
Scenarios drawn at random out of six
1 + 3
One correct answer, three distractors, no penalty
60 · 120 min
According to prep sites, not in the official guide

15 exam reflexes

Words in the question that point to the answer

guarantee · cannot be bypassed · regardless
Deterministic code: hook, check inside the tool, permissions. Never a more insistent prompt.
first step · most efficient first diagnostic
The cheapest fix that targets the cause: tool description, /memory.
root cause · directly addresses
Fix the cause (description, schema, decomposition), don't add a downstream safeguard.
structurally eliminates
Change the structure (consolidate, separate), don't mitigate (examples).
before adding infrastructure
Prompt-side answer: explicit criteria, examples.
generalize to novel patterns
Annotated examples that include the reasoning.
logs show · analysis shows · 94 % of…
The answer is in the stem's numbers: read them before the options.
blocking · users waiting · within 30 minutes
Synchronous API. Batch = up to 24 h, no SLA.
same session · verify your work
Self-review bias: independent instance, fresh context.
all · every · always · read all files · give all tools
Often a distractor: context or tool overload.
classifier · vector store · separate model · async layer
Often over-engineering compared with a simple fix.
confidence score (self-reported) · sentiment
Not a reliable signal without calibration; not an escalation trigger.

Pairs not to confuse

Two similar answers, decided by one detail of the question. Each number opens a quiz question that tests that detail.

Prompt, hook, or code in the tool?

If: 'Guarantee', 'cannot be bypassed', money, identity, compliance

Code: a blocking PreToolUse hook, a check inside the tool, a check after the loop

S1D1-06D1-09D3-01B5

If: Behavior to steer, no risk if it is occasionally missed

Prompt: explicit criteria, examples

S3D4-13D4-01

Split or consolidate tools?

If: Operations with different contracts (different required parameters, different outputs)

One tool per operation, each with its own schema

D2-01

If: Tools that overlap in meaning (credit vs refund, info already returned elsewhere)

Consolidate (action parameter, optional flag)

D2-02D2-12

Description, server-side safeguard, or routing?

If: 'First step', 'root cause', wrong tool choice or wrong parameter value

Rewrite the description (when to use it, where the value comes from)

S2D2-05D2-07

If: Server-side safeguard (validation, rejection)

A good complement, rarely the answer when the question asks for the cause

S2D2-05

Examples (few-shot) or explicit criteria?

If: 'Generalize to unseen cases', unstable format, ambiguous granularity

2 to 4 annotated, varied examples, with the reasoning

D4-04D4-03

If: Accurate but useless findings, 'before adding infrastructure'

Explicit criteria: what to report, what to ignore

D4-13S3

Plan mode or direct execution?

If: One file, clear change, bug located by a stack trace (even in an unfamiliar module)

Direct execution

D3-02

If: Several possible approaches, architecture decision, migration across dozens of files

Plan mode, then execution

S5

Escalate or resolve?

If: Explicit request for a human ('I want a real person NOW')

Immediate escalation, without investigating first

D5-06B23

If: Frustration without a clear request, problem solvable right away

Acknowledge, offer to resolve now, offer the human

D5-07

If: Sentiment, complexity, self-reported confidence

Not escalation triggers

B23S3

Resume, fork, or start over?

If: Context still valid, a few files changed

Resume + say precisely what changed

D1-11D3-13

If: Explore two directions from the same analysis

fork_session (independent branches)

D1-10

If: Stale tool results contaminating the answers

New session + structured summary + fresh calls

D5-12

Error or empty result?

If: Valid query, 0 results

Success with an empty list (isError false): don't retry

D2-06

If: Outage, timeout, access denied

isError true + category + isRetryable; protocol only for an unknown tool or an out-of-schema request

D2-10D2-09S8

Retry or not?

If: Validation error the model can fix

Retry with the precise error and the failed output

D4-09B9

If: Business error, timeout on a non-idempotent action, info missing from the source

No retry: explain, report the unknown status, or escalate

D2-09D2-11D4-05

Batch or synchronous API?

If: Someone is waiting, blocking check, alert within minutes

Synchronous

S11B12

If: Overnight report, results acceptable up to 24 h later

Batch (-50%), custom_id to join results

S11D4-11B10B11

CLAUDE.md, rules, @import, command, or settings?

If: Convention by file type or path (including scattered files)

.claude/rules/ with paths:

S6B18

If: Only the maintainers know which standards apply

Selective @import in each CLAUDE.md

D3-08

If: Occasional procedure (checklist, workflow)

Committed slash command / skill; a strict prohibition = permissions.deny

D3-11S4D3-09B16

Sliding window or structured facts?

If: The data shows old context is rarely used and easy to restate

Simple sliding window

D5-10

If: Several topics with IDs, amounts, statuses to recall later

A structured facts block outside the summary

D5-09D5-11

Fixed workflow or dynamic orchestrator?

If: Same steps, known in advance, for every input

Prompt chaining / coded workflow

D1-12S12

If: Subtasks that depend on what is discovered

Adaptive decomposition, orchestrator

D1-13D1-15

Required or nullable field?

If: The info may be missing from the source

Optional / nullable + an instruction to return null

D4-05

If: The info is in the source but in varied forms

Varied examples (don't make the field optional)

D4-03

Restrict or pre-approve?

If: Remove tools from an agent or a skill

tools / disallowedTools (AgentDefinition), disallowed-tools (skill)

B4S9

If: Avoid the permission prompt for a few tools

allowed-tools (skill, current turn), --allowedTools (CLI)

B14

Tool available or tool loaded in context?

If: 'How does the agent access tools across servers?'

All tools from all servers, discovered at connection time, available together

D2-17

If: What actually takes up context in Claude Code

With tool search: names only; description and schema loaded on demand

D2-17S9

The 6 scenarios

Customer Support
Support agent (Agent SDK) with MCP tools get_customer, lookup_order, process_refund, escalate_to_human. Goal: 80%+ first-contact resolution, while knowing when to escalate.
Claude Code
A team using Claude Code (generation, refactoring, debugging, documentation): slash commands, CLAUDE.md, plan mode vs direct execution.
Multi-Agent Research
Coordinator + subagents (web search, document analysis, synthesis, report) that produce cited reports.
Developer Productivity
Agent (Agent SDK) that helps explore codebases, understand legacy code and generate boilerplate; tools Read, Write, Bash, Grep, Glob + MCP servers.
CI/CD
Claude Code in CI: automated reviews, test generation, PR feedback; actionable feedback, few false positives.
Data Extraction
Structured extraction from documents, validated with JSON Schema, robust to edge cases and integrated with downstream systems.