中文版本

A curated resource list and study kit for Claude Certified Architect - Foundations (CCA-F).

This repository collects the best public resources for CCA-F preparation: official Anthropic materials, domain guides, hands-on labs, practice questions, open-source study projects, third-party references, and community exam reports. It also includes a local curriculum for learners and instructors who want a structured path instead of a loose bookmark list.

Request access to Claude Certified Architect - Foundations

• What This Is
• Start Here
• Exam At A Glance
• Repository Study Kit
• Official Anthropic Resources
• Domain Resources
• Open-Source Study Projects
• Practice Questions And Mock Exams
• Claude Code And Agent Engineering
• Third-Party Guides And Study Plans
• Community Reports
• Study Paths
• How To Use External Resources
• Maintenance Principles

CCA-F is not mainly a memorization exam. It tests architecture judgment around Claude systems: agent orchestration, tool design, Claude Code workflows, structured output, context management, reliability, and human escalation.

This project is organized like an awesome list, with one difference: it also contains original local study material under domains/, labs/, practice/, and reference/.

If you only have limited time, use these first.

If Anthropic updates the exam guide, access policy, scoring, or domain weights, follow the current official guidance.

Domain 1: Agentic Architecture & Orchestration    ██████████████ 27%
Domain 2: Tool Design & MCP Integration           █████████      18%
Domain 3: Claude Code Configuration & Workflows   ██████████     20%
Domain 4: Prompt Engineering & Structured Output  ██████████     20%
Domain 5: Context Management & Reliability        ███████        15%

Recommended study priority: D1 > D3 = D4 > D2 > D5.

See exam scenario notes for the local breakdown.

This repository is more than a link list. It contains a complete local study kit that mirrors the five official domains.

CCA-F/
├── README.md                         # Awesome CCA-F default README
├── README.zh.md                      # Chinese version
├── assets/                           # README images
├── domains/                          # Five domain deep dives
├── labs/                             # Jupyter notebooks
├── practice/                         # Practice questions and samples
└── reference/                        # Scenarios, tips, and glossary

pip install anthropic python-dotenv jupyter
export ANTHROPIC_API_KEY="your_api_key_here"
jupyter notebook

These are not CCA-F-specific in every case, but they are useful for the engineering concepts tested by the exam.

Community posts are useful for exam feel, but they are not authoritative. Use them to identify blind spots and then verify against official material.

1. Day 1: Map D1 and D3 core mechanisms.
2. Day 2: Review D4, especially schema, few-shot, temperature, and batch.
3. Day 3: Review high-frequency traps in D2 and D5.
4. Day 4-5: Practice by domain and tag mistakes by task statement.
5. Day 6: Rework official samples and all six scenarios.
6. Day 7: Review only mistakes, glossary, and cheat sheets.

## Question / Source

- Domain:
- Signal words:
- My answer:
- Correct answer:
- Mistake tag:
- Why the correct answer is more engineering-oriented:
- What signal should I notice next time:

Common mistake tags:

1. Confirm scope with the official exam guide and Anthropic Docs.
2. Use this repository's domain guides and labs to build a working structure.
3. Use open-source guides to compare explanations and fill gaps.
4. Use mock exams to train scenario judgment, not to memorize answers.
5. Treat third-party answers as hypotheses until verified against official docs.
6. Use community reports to discover blind spots, not as factual authority.

Core answer-selection heuristics:

High-risk rule      -> programmatic enforcement before prompt guidance
Wrong tool choice   -> improve tool name, description, schema, and boundaries
Agent loop issue    -> inspect stop_reason and tool_result handling
Subagent gap        -> explicitly pass context and provenance
CI workflow         -> use non-interactive mode and structured output
Extraction failure  -> schema + validation + retry
Long context        -> preserve facts, state, source, and uncertainty

• Keep this project as a curated Awesome CCA-F resource list and local study kit.
• Separate official, local, open-source, third-party, and community resources.
• Prefer resources that improve architecture judgment, not just answer volume.
• When official materials change, update exam facts and domain mappings first.
• Do not present third-party questions, blogs, or Reddit posts as official guidance.