▲ Claude Academy
Sign in

Vault / raw/2026-05-28-cca-f-overview.md

CCA-F overview (raw)

Exam basics

  • Name: Claude Certified Architect — Foundations (CCA-F)
  • Launched: 2026-03-12. Anthropic's first official technical certification.
  • Length: ~120 minutes, proctored, 60 multiple-choice questions, hosted on Skilljar.
  • Recommended experience: 6+ months hands-on with the Claude API, Claude Code, Agent SDK, and MCP.
  • Question style: 1 correct + 3 distractors. 4 scenarios per test, randomly sampled from a 6-scenario pool.

5 Domains (with weights)

  1. Agentic Architecture & Orchestration — 27% — multi-agent coordination, task decomposition, session state.
  2. Tool Design & MCP Integration — 18% — tool descriptions, error responses, MCP servers.
  3. Claude Code Configuration & Workflows — 20% — CLAUDE.md hierarchy, custom skills, CI/CD.
  4. Prompt Engineering & Structured Output — 20% — few-shot, JSON schemas, batches, validation loops.
  5. Context Management & Reliability — 15% — long conversations, escalation, error handling, CALM framework, cache_control breakpoints, conversation compaction, token budgets.

6 Scenarios (test pool)

  1. Customer Support Resolution Agent
  2. Code Generation with Claude Code
  3. Multi-Agent Research System
  4. Developer Productivity Tools
  5. Claude Code in CI/CD
  6. Structured Data Extraction

Learning-path courses

  • Claude 101 (beginner)
  • Building with the Claude API (most comprehensive cert-prep course)
  • Claude in Amazon Bedrock
  • Introduction to Model Context Protocol
  • Claude Code in Action
  • Introduction to Agent Skills
  • Introduction to Subagents
  • Model Context Protocol: Advanced Topics

Courses are free on Anthropic Academy (Skilljar). Launched 2026-03-02.

11 community study domains (daronyondem study guide)

  1. API fundamentals & output control (stateless requests, tool_choice, structured outputs, response prefill)
  2. Tool interface design for LLM agents (parameter design, structured output, composition, confirmation)
  3. Error handling in agent tools (transient vs permanent, structured errors, uncertain state, MCP patterns)
  4. Structured data extraction & validation (schemas, nullability, semantic validation, provenance, human review)
  5. Conversation context management (progressive summarization, state objects, retrieval, stale data)
  6. System prompt engineering (structure, principles vs conditionals, dilution, few-shot)
  7. MCP (tools, resources, prompts, trust model, descriptions, scopes)
  8. Agentic patterns (prompt chaining, routing, orchestrator-workers, subagents, provenance)
  9. Customer service workflow (escalation, compliance, graceful degradation, handoff)
  10. Claude Code & Agent SDK (built-in tools, plan mode, sessions, memory, slash commands, hooks)
  11. Evaluation & batch processing (validation, false-positive reduction, Message Batches, cost/latency)