Claude Certified Architect – Foundations
Certification Covered
CCA-F Certification Training
Stack
Program Includes:
Description
Claude Certified Architect – Foundations (CCA-F) is a 40-hour, instructor-led program for software engineers, AI engineers and solution architects who build production applications with Claude. Aligned to the official exam guide, it starts with the essentials of the Claude API, Claude Code and MCP, then builds the judgement the exam tests: designing agentic loops and multi-agent systems with the Claude Agent SDK, writing tools Claude selects reliably, configuring Claude Code for teams, producing dependable structured output, and keeping long-running agents accurate. You will apply these skills across the agent development lifecycle, from planning and evaluation to CI/CD pipelines and production monitoring. The program includes 80+ hands-on labs, scenario drills and a practice exam, and ends with two scenario-based capstone projects.
Course Outcomes
Upon completion, participants will be able to:
- Design agentic loops and coordinator–subagent systems with the Claude Agent SDK, enforcing critical rules through hooks rather than prompts
- Build MCP tools with clear descriptions, structured errors and scoped tool access, using the right tool_choice setting
- Configure Claude Code for teams with the CLAUDE.md hierarchy, path-specific rules, slash commands, skills and plan mode
- Produce reliable structured output with tool use, JSON schemas, few-shot examples and validation–retry loops
- Manage context, escalation and error propagation in long-running and multi-agent workflows while preserving provenance
- Apply the agent development lifecycle (ADLC) and run Claude Code in CI/CD for PR review, test generation and release gates
Prerequisites
- Programming: Working knowledge of Python or JavaScript/TypeScript (Python is used in most labs).
- APIs and JSON: REST APIs, JSON and JSON Schema basics.
- Developer tooling: Command line, Git and GitHub, and VS Code or a similar editor.
- CI/CD concepts: Basic understanding of pipelines, jobs and triggers (for example, GitHub Actions).
- LLM fundamentals: Prompts, tokens and context windows at a general level. Claude API, Agent SDK, Claude Code and MCP basics are covered in Module 0.
- Lab environment: An Anthropic API key, Claude Code installed, a Python or Node.js environment, and a GitHub repository with Actions enabled.
Course Content
1Foundations: Claude API, Prompting, Claude Code & MCP2 Sections
1.1 - Claude API & Prompting Essentials
1.1 - Claude API & Prompting Essentials
1.2 - Claude Code, Agent SDK & MCP Essentials
1.2 - Claude Code, Agent SDK & MCP Essentials
2Agentic Architecture & Orchestration7 Sections
2.1 - Agentic Loop Design with stop_reason
2.1 - Agentic Loop Design with stop_reason
2.2 - Coordinator - Subagent Orchestration
2.2 - Coordinator - Subagent Orchestration
2.3 - Subagent Invocation & Context Passing
2.3 - Subagent Invocation & Context Passing
2.4 - Workflow Enforcement & Human Handoffs
2.4 - Workflow Enforcement & Human Handoffs
2.5 - Agent SDK Hooks for Interception & Normalization
2.5 - Agent SDK Hooks for Interception & Normalization
2.6 - Task Decomposition Strategies
2.6 - Task Decomposition Strategies
2.7 - Session Resumption & Forking
2.7 - Session Resumption & Forking
3Tool Design & MCP Integration5 Sections
3.1 - Tool Interface Design & Descriptions
3.1 - Tool Interface Design & Descriptions
3.2 - Structured Error Handling in MCP Tools
3.2 - Structured Error Handling in MCP Tools
3.3 - Tool Distribution & tool_choice
3.3 - Tool Distribution & tool_choice
3.4 - MCP Servers in Claude Code & Agents
3.4 - MCP Servers in Claude Code & Agents
3.5 - Built-in Tools: Read, Write, Edit, Bash, Grep, Glob
3.5 - Built-in Tools: Read, Write, Edit, Bash, Grep, Glob
4Claude Code Configuration & Workflows6 Sections
4.1 - CLAUDE.md Hierarchy & Modular Configuration
4.1 - CLAUDE.md Hierarchy & Modular Configuration
4.2 - Custom Slash Commands & Skills
4.2 - Custom Slash Commands & Skills
4.3 - Path-Specific Rules
4.3 - Path-Specific Rules
4.4 - Plan Mode vs Direct Execution
4.4 - Plan Mode vs Direct Execution
4.5 - Iterative Refinement Techniques
4.5 - Iterative Refinement Techniques
4.6 - Claude Code in CI/CD Pipelines
4.6 - Claude Code in CI/CD Pipelines
5Prompt Engineering & Structured Output6 Sections
5.1 - Explicit Criteria to Reduce False Positives
5.1 - Explicit Criteria to Reduce False Positives
5.2 - Few-Shot Prompting for Consistency
5.2 - Few-Shot Prompting for Consistency
5.3 - Structured Output with tool_use & JSON Schema
5.3 - Structured Output with tool_use & JSON Schema
5.4 - Validation, Retry & Feedback Loops
5.4 - Validation, Retry & Feedback Loops
5.5 - Batch Processing with the Message Batches API
5.5 - Batch Processing with the Message Batches API
5.6 - Multi-Instance & Multi-Pass Review
5.6 - Multi-Instance & Multi-Pass Review
6Context Management & Reliability6 Sections
6.1 - Preserving Context in Long Interactions
6.1 - Preserving Context in Long Interactions
6.2 - Escalation & Ambiguity Resolution
6.2 - Escalation & Ambiguity Resolution
6.3 - Error Propagation in Multi-Agent Systems
6.3 - Error Propagation in Multi-Agent Systems
6.4 - Context Management for Large Codebases
6.4 - Context Management for Large Codebases
6.5 - Human Review & Confidence Calibration
6.5 - Human Review & Confidence Calibration
6.6 - Provenance & Uncertainty in Synthesis
6.6 - Provenance & Uncertainty in Synthesis
7Applied ADLC: Agent Development Lifecycle and CI/CD with Claude4 Sections
7.1 - The Agent Development Lifecycle (ADLC) with Claude
7.1 - The Agent Development Lifecycle (ADLC) with Claude
7.2 - Building CI/CD Pipelines with Claude Code
7.2 - Building CI/CD Pipelines with Claude Code
7.3 - Evaluation & Release Gates for Agents
7.3 - Evaluation & Release Gates for Agents
7.4 - Operating & Improving Agents in Production
7.4 - Operating & Improving Agents in Production
8Certification Readiness & Practice Exam1 Sections
8.1 - Scenario Drills, Practice Exam & Exam Strategy
8.1 - Scenario Drills, Practice Exam & Exam Strategy
Overview of Capstone Projects
Customer Support Resolution Agent
Customer Support Resolution Agent
- Objective: Build a support agent with the Claude Agent SDK that resolves returns, billing disputes and account issues through custom MCP tools (get_customer, lookup_order, process_refund, escalate_to_human), targeting 80%+ first-contact resolution while escalating the right cases.
- What you will build:
- Agentic loop and orchestration driven by stop_reason, with multi-concern requests decomposed and investigated in parallel.
- MCP tools with clearly differentiated descriptions and structured errors (errorCategory, isRetryable).
- Deterministic enforcement: a prerequisite gate that blocks refunds until identity is verified, and a PreToolUse hook that redirects refunds above $500 to a human.
- PostToolUse normalization of dates and status codes coming from different tools.
- Context and escalation: a persistent "case facts" block, trimmed tool outputs, explicit escalation criteria with few-shot examples, and structured handoff summaries.
- ADLC: a golden dataset of 40+ conversations (including policy gaps and multiple customer matches) run as a regression suite in CI.
- Deliverables:
- Agent repository — agentic loop, MCP servers, hooks, prerequisite gates and subagent definitions.
- Golden dataset and results — 40+ conversations with pass/fail results by scenario type.
- Escalation design note — escalation criteria, few-shot examples and a sample structured handoff.
Tech Stack
Claude-Powered CI/CD Quality Pipeline
Claude-Powered CI/CD Quality Pipeline
- Objective: Configure Claude Code for a team repository and wire it into a CI/CD pipeline that reviews pull requests, generates tests and produces a nightly technical-debt report, with actionable feedback and few false positives.
- What you will build:
- Team configuration: project CLAUDE.md with @import, path-specific rules in .claude/rules/, a shared /review command, a skill using context: fork and allowed-tools, and a shared MCP server in .mcp.json.
- Pre-merge review with claude -p, --output-format json and --json-schema, posting findings as inline PR comments and reporting only new issues on re-runs.
- Multi-pass review: per-file passes plus a cross-file integration pass, run by an independent instance rather than the generating session.
- Precision tuning with explicit report/skip criteria, severity examples and a detected_pattern field for analyzing dismissed findings.
- Test generation that uses documented fixtures in CLAUDE.md and existing tests to avoid duplicates.
- Nightly batch job on the Message Batches API, with failures resubmitted by custom_id.
- Deliverables:
- Team repository — CLAUDE.md hierarchy, .claude/ rules, commands and skills, .mcp.json and the code from both capstones.
- CI/CD workflows — PR review, test generation and the nightly batch job, with an evaluation gate.
- Evaluation report — golden-dataset results broken down by scenario type.
- Architecture Decision Record — 500–800 words justifying design choices and trade-offs across all five domains.
- Evaluation Rubric (aligned to exam weightage):
- Agentic architecture and orchestration (27%) — loop control, subagent design, enforcement, hooks and session handling.
- Tool design and MCP integration (18%) — descriptions, structured errors, scoped tools and MCP configuration.
- Claude Code configuration and workflows (20%) — CLAUDE.md, rules, commands, skills and CI/CD integration.
- Prompt engineering and structured output (20%) — criteria, few-shot examples, schemas, retry loops and batch design.
- Context management and reliability (15%) — context handling, escalation, error propagation and provenance.
Tech Stack
Career Opportunities
Learning Roadmap
Foundations: Claude API, Prompting, Claude Code & MCP
2 Sections
Agentic Architecture & Orchestration
7 Sections
Tool Design & MCP Integration
5 Sections
Claude Code Configuration & Workflows
6 Sections
Prompt Engineering & Structured Output
6 Sections
Context Management & Reliability
6 Sections
Applied ADLC: Agent Development Lifecycle and CI/CD with Claude
4 Sections
Certification Readiness & Practice Exam
1 Sections
Certified Claude Certified Architect – Foundations
Goal Achieved
Weekdays (Mon-Fri)
Starting from 16th November
Duration: 2 Hours per day
Weekend (Sat & Sun)
Starting from 21st November
Duration: 4 Hours per day
Request For Demo
How do I enroll in this course?
Explore
Explore the course and certification
Consult
Talk to our agents for guidance
Experience
Join our live demo session
Enroll
Complete your enrollment
Meet Our Leaders
Program Director
Set the direction of the program and ensure it meets industry and academic excellence standards.
Aashish Pandey
Blue Data Consulting
Program Manager
Manages schedules, coordination, assessments, and ensures smooth course delivery.
Ragini Sankhwar
Blue Data Consulting
Industry Expert/Trainers
Harsh Dalal
Blue Data Consulting
AI Consultant
SME
Covers theory, hands-on labs, assessments, and daily assignments.
Aproova
Blue Data Consulting
Master SME
Specialized guidance for Capstone Projects and advanced implementation.
Ahmed
Blue Data Consulting
Mentor
Dedicated support to help all learners pass the certification exam.
Transparent Pricing
Invest in your future with our flexible payment options
Application Fee
5% of Total Fee (Non-refundable)
Program Fee
Complete program access
(After Application Fee + 20% Discount)
*18% GST & Certification cost included
*Important: First pay Application Fee, then you can enroll in Program
Easy EMI Plans
Pay in flexible monthly installments
Starting from
Frequently Asked Questions
What is the CCA-F certification?
The Claude Certified Architect – Foundations (CCA-F) certification training is a comprehensive 32-hour program designed to validate deep, hands-on proficiency with the Anthropic Claude ecosystem.
What core domains are covered in the training?
The training covers five core domains:
- Agentic architecture
- Tool and MCP integration
- Claude Code configuration
- Prompt engineering
- Context management
What is the format of the exam?
The exam consists of multiple-choice questions. Each question features one correct answer and three distractors. There is no penalty for guessing.
What score do I need to pass?
Candidates must achieve a scaled score of 720 out of 1,000 to pass the certification exam.
Who is the target audience for this certification?
The target audience includes:
- Software engineers, AI engineers, and solution architects with 6 or more months of hands-on experience building with Claude APIs, Claude Agent SDK, Claude Code, and MCP integrations.
- Technical leads and platform engineers responsible for designing and deploying production-grade agentic systems, multi-agent workflows, and Claude-powered developer tooling.
- DevOps and MLOps engineers integrating Claude Code into CI/CD pipelines and seeking to validate advanced configuration, prompt engineering, and context management proficiency.
What are the prerequisites to take this course?
The prerequisites include:
- Familiarity with REST APIs and JSON/JSON Schema.
- Working knowledge of at least one programming language, with Python or JavaScript/TypeScript recommended.
- A basic understanding of CI/CD pipeline concepts.
- Familiarity with command-line tools and version control (Git).
- Understanding of LLM fundamentals including prompting, context windows, and tokens.
What are the technical lab requirements?
To participate in the labs, you will need:
- An Anthropic API key with access to Claude models, specifically claude-sonnet-4-6 recommended.
- Claude Code CLI installed and configured.
- A Node.js or Python development environment.
- A Git repository for Claude Code configuration exercises.
- At least one MCP server for integration practice.
- A CI/CD pipeline environment for Domain 3 labs.
- A code editor with terminal access.
- Access to Anthropic documentation at docs.claude.com.
How many scenarios are tested on the exam?
There are 6 total scenarios available, but you will only be tested on 4 scenario sets presented at random during the exam.
What is the capstone project?
The capstone project requires candidates to build an Intelligent Research & Report Generation System. You must design and implement a multi-agent research system that accepts a topic, autonomously decomposes it, retrieves information via MCP tools, and produces a structured, citation-backed report.
How is the capstone project evaluated?
The evaluation rubric is broken down into five categories:
- Agentic Architecture accounts for 25%.
- Tool Design & MCP Integration accounts for 20%.
- Claude Code Configuration accounts for 20%.
- Prompt Engineering & Structured Output accounts for 20%.
- Context Management & Reliability accounts for 15%.
What deliverables are expected for the capstone?
Candidates are expected to deliver:
- A source code repository containing the full multi-agent system, CLAUDE.md hierarchy, .mcp.json, custom slash commands, and Skill definitions.
- A completed sample report generated by the system for a provided test research topic.
- CI/CD pipeline configuration demonstrating non-interactive Claude Code execution.
- A written Architecture Decision Record (ADR) between 500 and 800 words justifying key design choices.
- A peer review checklist completed for one other candidate’s submission.
Where can I request exam access or view reference materials?
You can request official access to the training at anthropic.skilljar.com/claude-certified-architect-foundations-access-request. For reference materials, you can visit the Anthropic Documentation at docs.claude.com.