Skip to content

Exam Overview & Blueprint

Review the formal structure, passing thresholds, tested scenarios, and content domains for the Devs And Agents study program and the CCA-F exam.

📊 Exam at a Glance

Quick overview of candidate profile, format, and passing score.

  • Target Candidate

    Solution architects and developers with 6+ months of experience building with Claude APIs, Agent SDK, Claude Code, and MCP.

  • Exam Format

    Multiple choice, single correct answer. Grounded in realistic customer production scenarios. No penalty for guessing.

  • Thresholds

    Scaled score between 100 – 1000. The minimum passing score is 720.

🎯 The Five Content Domains

Select a domain to begin studying. Each card links directly to the first module in that domain.

  • Domain 1: Agentic Architecture & Orchestration

    Weight: 27%

    Focuses on:

    • Core agentic loops & stop_reason control flow
    • Coordinator-subagent hub-and-spoke patterns
    • Hooks (PostToolUse, prerequisite gates)
    • Task decomposition & session management

    Explore Domain 1 →

  • Domain 2: Tool Design & MCP Integration

    Weight: 18%

    Focuses on:

    • Clear tool descriptions & boundary definitions
    • Structured error handling & metadata
    • Scoping tools & using tool_choice
    • Exposing resources & configuring .mcp.json

    Explore Domain 2 →

  • Domain 3: Claude Code Configuration & Workflows

    Weight: 20%

    Focuses on:

    • Scoping and modularizing CLAUDE.md files
    • Custom slash commands and Agent Skills
    • Conditional loaded path-specific rules
    • Determining Plan mode vs Direct execution

    Explore Domain 3 →

  • Domain 4: Prompt Engineering & Structured Output

    Weight: 20%

    Focuses on:

    • Writing prompts with explicit criteria
    • Targeted few-shot prompting techniques
    • Enforcing JSON schemas via tool use
    • Designing validation & batch API runs

    Explore Domain 4 →

  • Domain 5: Context Management & Reliability

    Weight: 15%

    Focuses on:

    • Fact extraction & token budget management
    • Designing human-in-the-loop workflows
    • Error propagation & coordinator recovery
    • Synthesizing conflicts & source provenance

    Explore Domain 5 →

🎬 Tested Production Scenarios

The exam questions are framed around these six core production use cases.

  • 1. Customer Support Agent

    Custom Model Context Protocol (MCP) billing/return tools and human escalation rules.

  • 2. Claude Code Generation

    CLAUDE.md hierarchies, custom slash commands, and plan vs. direct mode.

  • 3. Multi-Agent Systems

    Spawning subagents using the Task tool and passing structured search results.

  • 4. Developer Productivity

    Codebase exploration using built-in file tools (Read, Write, Grep, Glob).

  • 5. Continuous Integration

    Non-interactive automated reviews with JSON outputs in CI pipelines.

  • 6. Structured Data Extraction

    Schema enforcement via tool use, validation-retry loops, and batch API processing.