Claude Certified Architect – Foundations (CCAR-F)

AI-assisted development creates governance questions for IT leaders: teams need the productivity benefits of Claude Code and Claude APIs without turning one-off prompts into unreviewable production risk. This 40-skill, 32.4-hour intermediate course is a practical fit for software engineers, backend developers, system architects, and AI-focused IT Practitioners who already work with APIs and code workflows.

IT teams complete this training to standardize how they set up Claude Code, manage project context, plan code changes, build repeatable commands, evaluate prompts, use tools and RAG, and design agent workflows with observable traces. The course repeatedly frames reliability around scoped tasks, success criteria, debugging, evaluation, human approval, and auditability — useful change-management anchors for Team Leads adopting AI coding assistants across a team.

For rollout, Training Managers can assign the course in sections over multiple weeks. CBT Nuggets Playlists can sequence the 40 skills by role or project phase, while Team Reporting helps IT Directors track progress across assigned learners.

This training helps teams move from AI experimentation to safer, repeatable Claude-based development workflows. The course uses concrete build scenarios including a weather app, custom Claude Code commands, tool-driven workflows, RAG and vector search, multimodal inputs, and a capstone agentic multimedia article system with evaluation, repair, packaging, and approval.

• Standardize AI-assisted code changes: Teams learn to use planning modes, success criteria, context controls, screenshots, interrupt/rewind, and memory so Claude Code work stays scoped and reviewable.
• Reduce inconsistent prompting: Custom commands turn common work — dependency audits, test writing, pull request summaries, convention checks — into repeatable workflows instead of ad hoc prompts.
• Improve production readiness: Prompt evaluation, test datasets, graders, repair attempts, and reruns help teams detect defects before release.
• Support review and incident analysis: The capstone emphasizes observable boundaries, checksums, human approval, and traceability from request to release.

Knowledge

• How Claude Code fits into CLI, IDE terminal, and plugin-based development workflows.
• How Claude API foundations include structured messages, models, controls, streaming, prompts, stop sequences, system prompts, and temperature.
• How prompt design, examples, XML tags, test datasets, and graders support evaluation.
• How tool use, tool schemas, tool calls, tool results, and multi-turn tool conversations enable agent behavior.
• How RAG, document chunking, embeddings, vector spaces, vector search, and hybrid retrieval support grounded responses.
• How auditability, observability, human approval, and repair loops reduce risk in agentic systems.

Ability

• Set up Claude Code projects and manage project context for focused code changes.
• Create custom commands that make team workflows repeatable and easier to review.
• Build and route tool-based Claude workflows, including safer use of built-in tools.
• Apply image, PDF, caching, code execution, and file storage capabilities in Claude applications.
• Build MCP tools, resources, and prompts, then test MCP systems with the Inspector.
• Design reusable agent skills and assemble a skill-aware agent workflow.