Scenarios taught as production architectures.
14 reference scenarios covering roughly 60% of the CCA-F (Claude Certified Architect - Foundations) exam value, plus question-driven deep dives on common production failure patterns. Each scenario is taught as a production architecture: 6 colored diagrams, 4-row decision matrix, failure-mode catalog, an 8-step build, and exam-pattern questions. Concepts answer what is X; scenarios answer how do I build X.
Start here
Customer support resolution: the single highest-weight scenario on the exam, three domains, one architecture.
Start here.
The single highest-weight scenario on the exam, three domains (D1 + D2 + D5), one architecture.
Customer Support Resolution Agent
A production refund/escalation agent built on Claude. The harness reads stop_reason, dispatches tools through a registry, gates risky operations with a PreToolUse hook (deterministic policy enforcement), pins customer state in a case-facts block, and routes blocked calls to a structured escalation queue. Three-domain coverage makes this the single highest-weight scenario on the exam.
Pick your weakest area first.
Live scenarios are clickable; planned ones list their target quarter. The catalog is intentionally short - 14 architectures that, between them, exercise every exam concept, plus deep dives for question-driven failure patterns.
D1 Agentic Architectures
27%Customer Support Resolution Agent
A production refund/escalation agent built on Claude. The harness reads stop_reason, dispatches tools through a registry, gates risky operations with a PreToolUse hook (deterministic policy enforcement), pins customer st
Multi-Agent Research System
A hub-and-spoke research system. The coordinator owns task decomposition (semantic, not lexical), spawns 3-5 research subagents in parallel with isolated contexts and scoped tools, routes findings through a verification
Developer Productivity Agent
A team of specialised subagents that handle codebase exploration, code review, and doc generation. Each with a narrow tool whitelist and isolated context. The lead agent uses Grep + Glob first to locate relevant files (n
Claude for Operations
An on-call ops agent. Alerts arrive structured (service, severity, root_cause_signal, 4-bucket isError: Transient · Permission · Data · Business); the agent matches a runbook from a small registry (3-5 safe playbooks: re
D2 Tool Design + Integration
18%Structured Data Extraction
A schema-driven extraction agent. The harness defines the output shape as a tool with input_schema, sets tool_choice to force that tool, runs a validation-retry loop (parse → validate → on failure feed the error back), u
Agentic Tool Design
A meta-skill scenario about designing tool registries. The optimum is 4-5 tools per agent (past 5, routing accuracy drops 8% per extra tool); each tool follows the Anthropic 4-line description pattern (what / when / edge
Agent Skills with Code Execution
A code-execution Skill with four layers of safety. Layer 1: route file I/O to built-in tools (Read, Write, Edit) and reserve Bash only for actual execution. Layer 2: a PreToolUse hook scans the proposed Bash command for
Invoice Processing Agent
An AP-automation agent that wraps four guarantees around invoice approval. (1) Forced tool_use with a strict JSON schema (vendor_id, invoice_number, line_items[], total_amount, currency ISO 4217, due_date ISO 8601, PO_re
Tool-Call Skipping: When Claude Bypasses a Required Tool
Why Claude agents skip required tool calls and how to enforce them. Index of deep-dives on tool-call skipping with the canonical 12% baseline and the structured fixes.
When Your Agent Bypasses get_customer and Calls lookup_order Directly
Diagnostic guide for a 12% tool-skip rate where Claude calls lookup_order without first calling get_customer - schema-level fixes, prompt rails, and CCA-F D2/D3 exam mapping.
D3 Agent Operations
20%Code Generation with Claude Code
Claude Code as a team-aware code-generation agent. The repo carries a CLAUDE.md hierarchy (project + personal + system) committed to version control, area-specific rules in .claude/rules/ keyed by file glob, Plan Mode as
Claude Code for CI/CD
Claude Code as a headless PR reviewer in GitHub Actions. The workflow uses claude -p for non-interactive execution, runs per-file independent sessions (no shared context across the 14 files in a PR. That prevents lost-in
Agent Skills for Enterprise KM
An enterprise-scale Skills registry. Each Skill is a markdown file with frontmatter (name + version + description + tags + dependencies + access_level), stored in .claude/skills/{team}/{name}.md so naming collisions beco
Agent Skills for Developer Tooling
A CLI-first Skills architecture for developer tooling. Each Skill is a markdown file with frontmatter (name, version, parameters, allowed-tools); the CLI invokes them via claude skills invoke <skill> --param key=value; r
D5 Context + Reliability
15%Conversational AI Patterns
A multi-turn conversational pattern that survives context compression. The harness pins a CASE_FACTS block at the top of every system-prompt iteration (immutable, re-read every turn), summarizes turns 2-14 into 3 lines a
Long Document Processing
A retrieval-augmented long-document agent. Semantic chunking (paragraph- and section-aware, not fixed-size) preserves meaning; an embedding index supports top-K retrieval so only the relevant chunks enter context; an imm
All 16 entries.
Status, primary domain, build time, exam coverage. Live rows link to the full scenario page.
| # | Title | Domains | Build | Exam weight | Status |
|---|---|---|---|---|---|
| P3.1 | Customer Support Resolution Agent | D1 + D2 + D5 | 22 min | 60% of CCA-F (D1 + D2 + D5) | ● Live |
| P3.10 | Conversational AI Patterns | D5 + D4 | 24 min | 35% of CCA-F (D5 + D4) | ● Live |
| P3.2 | Code Generation with Claude Code | D3 + D2 | 24 min | 38% of CCA-F (D3 + D2) | ● Live |
| P3.3 | Multi-Agent Research System | D1 + D2 | 26 min | 45% of CCA-F (D1 + D2) | ● Live |
| P3.5 | Claude Code for CI/CD | D3 + D2 | 25 min | 38% of CCA-F (D3 + D2) | ● Live |
| P3.11 | Agent Skills for Enterprise KM | D3 + D2 | 26 min | 38% of CCA-F (D3 + D2) | ● Live |
| P3.4 | Developer Productivity Agent | D1 + D2 | 22 min | 45% of CCA-F (D1 + D2) | ● Live |
| P3.6 | Structured Data Extraction | D2 + D5 | 22 min | 33% of CCA-F (D2 + D5) | ● Live |
| P3.7 | Agentic Tool Design | D2 + D3 | 24 min | 38% of CCA-F (D2 + D3) | ● Live |
| P3.8 | Long Document Processing | D5 + D2 | 26 min | 33% of CCA-F (D5 + D2) | ● Live |
| P3.9 | Claude for Operations | D1 + D5 | 26 min | 42% of CCA-F (D1 + D5) | ● Live |
| P3.12 | Agent Skills for Developer Tooling | D3 + D2 | 22 min | 38% of CCA-F (D3 + D2) | ● Live |
| P3.13 | Agent Skills with Code Execution | D2 + D3 | 24 min | 38% of CCA-F (D2 + D3) | ● Live |
| P3.14 | Invoice Processing Agent | D2 + D5 | 26 min | 33% of CCA-F (D2 + D5) | ● Live |
| Deep dive | Tool-Call Skipping: When Claude Bypasses a Required Tool | D2 + D3 | 6 min | D2 (18%) + D3 (20%) | ● Live |
| Deep dive | When Your Agent Bypasses get_customer and Calls lookup_order Directly | D2 + D3 | 8 min | D2 (18%) + D3 (20%) | ● Live |
6 questions about the scenario pool
Every Q is phrased as a real Google search query. Answers cite the same evidence-tagged sources used elsewhere on the site.