Skip to content
Study sections

Study Sections

The CCA-F study material is broken into 12 sections so each one comfortably fits in a Claude Code conversation. When you start Mock Exam Mode, Claude reads exactly one of these per round — that’s how the questions stay specific, the distractors stay plausible, and the context budget stays sane.

Pick a section to skim, or jump into Section 1 — Exam Overview for orientation. To practice with Claude as your proctor, see the home page for the kickoff prompt.

1. Exam Overview
Why CCA-F exists, scoring, the 6 scenarios, and a study path mapped to domain weights.
2. Agentic Loops
Domain 1.1 — agentic loop control flow, stop_reason values, anti-patterns.
3. Multi-Agent Orchestration
Domains 1.2–1.4 — coordinator/subagent boundaries, Task tool, context passing.
4. Hooks, Decomposition & Sessions
Domains 1.5–1.7 — PostToolUse hooks, prompt chaining vs adaptive decomposition, fork_session.
5. Tool Design & Built-ins
Domains 2.1, 2.3, 2.5 — tool descriptions, tool_choice, Read/Write/Edit/Bash/Grep/Glob.
6. MCP Integration
Domains 2.2, 2.4 — structured error envelopes, .mcp.json scopes, resources vs tools.
7. CLAUDE.md & Rules
Domains 3.1, 3.3 — CLAUDE.md hierarchy, @imports, .claude/rules/ glob patterns.
8. Commands, Skills & Plan Mode
Domains 3.2, 3.4, 3.5 — slash commands, skills with context: fork, plan mode.
9. CI/CD Integration
Domain 3.6 — claude -p, –output-format json, prompt design for CI.
10. Prompt Engineering
Domains 4.1, 4.2, 4.4 — explicit criteria, few-shot, multi-pass review.
11. Structured Output & Batch
Domains 4.3, 4.5, 4.6 — JSON Schema, validation-retry loops, Message Batches API.
12. Context & Reliability
Domain 5 — case-facts, escalation triggers, structured errors, claim–source mapping.

How to use these on the exam

  1. Skim section 1 first — it sets up the exam structure, the 6 scenarios, and the domain weights. Everything else hangs off it.
  2. Drill the heavy domains first. D1 (Agentic Architecture) is 27% of the exam and shows up across three scenarios — sections 2, 3, 4 are your highest-leverage hours.
  3. Then D3 + D4 (20% each) — sections 7, 8, 9 (Claude Code) and sections 10, 11 (prompt engineering / structured output).
  4. Finish with D2 and D5 — sections 5, 6 (tools / MCP) and 12 (context, escalation, provenance — small domain but cross-cutting).
  5. Use Mock Exam Mode in Claude Code to drill each section with real-format questions until you can predict the distractor pattern before reading the options.
Last updated on