Claude Certified Developer - Foundations
Build, integrate, secure, test, and ship Claude applications, agents, workflows, tools, and MCP servers.
Software engineers, AI engineers, and technical builders with Python or TypeScript experience and working knowledge of REST APIs and command-line tools.
53
Questions
120 min
Time limit
$125
Exam fee
720 / 1000
Passing score
Multiple-choice and multiple-response
Format
Proctored online or at a test center
Delivery
12 months
Validity
v1.0 · July 2026
Guide
Exam domains
Weighted objectives straight from the exam guide.
Agents and Workflows
14.7%- Choose workflows or agents from task constraints
- Design manager and subagent hierarchies
- Build with the Claude Agent SDK, custom loops, managed agents, and hooks
- Apply tool-loop, memory, context, and framework patterns
Applications and Integration
33.1%- Translate business needs into functional and infrastructure requirements
- Apply system lifecycle practices
- Use messages, tools, streaming, vision, thinking, caching, batch, and third-party providers
- Apply REST, JSON, async, version-control, review, and refactoring practices
- Design instruction, schema, session, content-boundary, and plugin behavior
- Manage CLAUDE.md, settings, model, prompt, and plugin versions
Claude Code
3.1%- Operate Rules, Skills, Commands, Agents, memory, sessions, headless mode, streaming mode, auto-mode, CLAUDE.md, initialization, and settings
Eval, Testing, and Debugging
2.6%- Classify errors, select recovery, analyze traces, and isolate integration failures from model failures
Model Selection and Optimization
16.8%- Explain tokens, sampling, context, non-determinism, thinking, effort, and prompting modes
- Integrate SDK and transport fundamentals
- Balance model quality, latency, cost, capability, and version changes
- Budget tokens and use prompt caching
Prompt and Context Engineering
11%- Prevent context drift and bloat with pruning, compaction, and isolation
- Place clear instructions, examples, and constraints
- Sanitize input and iterate prompts
- Validate and defensively consume structured output
Security and Safety
8.1%- Mitigate prompt injection, jailbreaks, data leakage, and unsafe input
- Layer guardrails with least privilege
- Use hooks for deterministic safety controls
- Protect identity, secrets, keys, and access approvals
Tools and MCPs
10.6%- Design descriptions, schemas, errors, approvals, and tool sets
- Author and deploy MCP tools, resources, and prompts across transports
- Choose among built-in tools, custom tools, Skills, and MCP
Study path · 15 lessons
Go deeper in the main curriculum
The Agent Loop
Build the loop before using an SDK
Claude Agent SDK
Go deeper on the production SDK surface
Tool Schema Design
Design clear, testable tool contracts
Building an MCP Server
Implement the server boundary end to end
Prompt Injection Defense
Threat-model untrusted instructions
Eval-Driven Agent Development
Use trajectories and final state as release evidence