everything-claude-code github: CCDV-F Developer Exam Prep

The everything-claude-code github search reflects a specific developer instinct: pull the full capability surface, understand the system, then sit the exam. For CCDV-F candidates, that instinct maps directly to exam strategy. Claude Code's GitHub-connected features (Actions runners, code review agents, CI/CD hooks) appear across several of the eight exam domains. This guide maps that surface to the domain weights so you study what the exam actually scores.

What is the CCDV-F exam, and how is it structured?

The Claude Certified Developer, Foundations exam (official code CCDV-F) validates practical skills in building, integrating, and shipping Claude-powered applications. It costs $125 per attempt, runs for 120 minutes, and contains 53 items in multiple-choice and multiple-response formats. The passing score is 720 on a scale of 100 to 1000. The credential is valid for 12 months from the date it is awarded and is delivered online-proctored or at a test centre via Pearson VUE.

Unlike the architect track (CCAR-F), CCDV-F has no scenario bank. Every item is written directly against the skills in each domain, which makes domain weight the clearest signal for where to study.

The weights above are exact percentages sourced from the official CCDV-F exam guide published 8 July 2026. Domain 2 alone accounts for roughly 17 to 18 questions on a 53-item exam, more than any other domain by a substantial margin.

Why does everything-claude-code github point toward CCDV-F?

Claude Code is Anthropic's agentic coding tool that runs in the terminal and integrates with GitHub workflows. When developers search everything-claude-code github, they are typically looking for the full capability surface: how Claude Code reads and writes code, how it connects to GitHub Actions, how it handles pull requests, and how it chains tool calls across a CI pipeline.

That capability surface maps to Claude Code Configuration and Workflows (Domain 3, 3.1%) and to Applications and Integration (Domain 2, 33.1%), where the exam tests how you build, connect, and ship Claude-powered applications. Because Domain 3 carries only 3.1% of the exam, some candidates underweight it. That is a mistake: understanding Claude Code's GitHub integration patterns reinforces Domain 2 content, which carries more than ten times the weight.

Which CCDV-F domain deserves the most study time?

Domain 2, Applications and Integration, at 33.1%, deserves more prep time than any other. It covers how you build real Claude-powered applications: connecting Claude to databases and APIs, handling authentication, managing webhooks, integrating into existing codebases, and shipping reliable production pipelines.

Domain 2 questions typically present a real-world integration scenario and ask which approach is most reliable, most cost-effective, or most appropriate for the stated constraints. Common patterns include webhook-driven integrations, database-backed Claude applications, and multi-step pipelines where each step calls Claude with a refined context.

For GitHub-specific prep, focus on how Claude Code participates in CI/CD workflows. A typical integration step routes pull request events through a Claude-powered review:

# GitHub Actions step invoking Claude Code for automated review
- name: Claude Code review
  uses: anthropic-ai/claude-code-action@v1
  with:
    prompt: "Review this diff for correctness and security issues."
    model: "claude-sonnet-5"

The exam will not test API syntax verbatim, but it will ask you to choose the right integration pattern for a scenario: whether a synchronous review fits a CI gate, or whether an asynchronous batch pattern is more appropriate for a high-volume repository. That judgment lives squarely in Domain 2.

The Tool Design and MCP Integration domain complements this. Tool design decisions made at the application layer directly affect integration reliability. A poorly described tool leads to misrouting; a correctly scoped MCP server keeps Claude's actions predictable inside a CI environment.

What should you study for Model Selection and Optimization (Domain 5, 16.8%)?

Domain 5 is the second-largest domain by weight. It tests cost-aware decisions: which model to select for a given task, how to balance latency against capability, and how to apply prompt caching and batching to reduce spend per request.

The exam rewards candidates who reason about cost-performance trade-offs. A question might describe a high-volume document classification job and ask you to choose between Haiku 4.5 (fast, low cost), Sonnet 5 (balanced), and Opus 5 (highest capability). The correct answer depends on the stated accuracy requirement and volume, not on recency or raw benchmark position.

A common mistake in Domain 5 prep is studying which model is fastest or cheapest in isolation. The exam always frames model selection in context: a given scenario, a stated quality bar, a specific cost constraint. Study the trade-offs, not the spec sheet.

Key study areas for Domain 5:

• Matching model capability to task complexity
• Prompt caching mechanics and when they apply
• Batch API versus synchronous requests for throughput-sensitive workloads
• Token budgeting at the application layer

How do Agents and Workflows appear on CCDV-F (Domain 1, 14.7%)?

Domain 1 tests when to use a simple workflow (fixed steps, deterministic control flow) versus a full agent loop (dynamic tool selection, multi-step reasoning). The exam consistently rewards simpler, more auditable solutions when the task is well-defined, and reserves agent loops for tasks that genuinely require adaptive behaviour.

For GitHub-adjacent scenarios, a common question pattern involves a code review pipeline: should you build a fixed prompt chain that always runs the same checks, or an agent that decides which checks to run based on the diff? The exam rewards the fixed pipeline when the checks are stable and known in advance, and marks down unnecessary agent loops that introduce non-determinism without a corresponding benefit.

Agentic Architecture and Orchestration concepts inform Domain 1, even though the architect track is a separate exam. Developer candidates benefit from understanding parallel subagent spawning and coordinator responsibilities as background knowledge, since CCDV-F Domain 1 tests similar reasoning at a shallower depth.

What does Security and Safety (Domain 7, 8.1%) test?

Domain 7 focuses on two main threat patterns: prompt injection and tool containment. Prompt injection attacks arrive through untrusted input (a GitHub issue body, a pull request description, or a code comment) and attempt to redirect Claude's actions. The exam tests whether you can identify injection vectors and apply defences such as input sanitisation, restricted tool scopes, and human confirmation gates for high-risk actions.

Tool containment is the structural side of the same problem. When Claude Code has access to shell commands, file writes, or network calls, the exam asks you to choose the minimum necessary permission scope. Granting execute_shell unconditionally is a pattern the exam will mark wrong; scoping it to a specific working directory with an allowlist is the pattern it rewards.

How does Prompt and Context Engineering (Domain 6, 11.0%) appear on the exam?

Domain 6 covers prompt design and context management: knowing what to include, what to compress, and what to exclude when a long session approaches its limits. For developer candidates, the most common patterns involve choosing between zero-shot, one-shot, and few-shot prompting based on task complexity, and designing system prompts that constrain Claude's behaviour without over-specifying.

The Prompt Engineering and Structured Output concepts apply directly here. A fenced prompt structure for a code review task illustrates the core Domain 6 pattern:

<task>
Review the following Python function for correctness and security issues.
Return your findings as a JSON array. Each item must have "type", "line", and "detail" fields.
</task>

<code>
{{code}}
</code>

The XML tags structure the input; the output schema instruction constrains the format. That combination is a Domain 6 staple: structure the input, specify the output.

How do Tools and MCPs (Domain 8, 10.6%) fit into developer certification?

Domain 8 covers MCP server setup, tool description quality, and the decision of when to build a custom MCP server versus using an existing one. The MCP Scoping Hierarchy governs which tools are available to Claude in a given context (project-level, user-level, or machine-level), and the exam tests whether you can diagnose a scoping failure when a tool is absent where expected.

Tool description quality is a recurring theme across Domains 2 and 8. A vague description causes Claude to misroute tool calls; a description that duplicates system-prompt instructions creates conflicts. The exam rewards candidates who understand that the tool description is the primary selection mechanism, and that precision there is lower-effort and higher-leverage than iterating on the system prompt.

{
  "name": "create_pull_request",
  "description": "Creates a GitHub pull request from the current branch to main. Use only after all tests pass. Do not call for draft PRs.",
  "input_schema": {
    "type": "object",
    "properties": {
      "title": { "type": "string" },
      "body": { "type": "string" }
    },
    "required": ["title", "body"]
  }
}

The description above constrains behaviour through specificity: it names the target branch, states a prerequisite condition, and excludes a class of misuse. That level of precision is what the exam rewards.

How does AI Skill Certs support CCDV-F preparation?

AI Skill Certs is an independent adaptive prep platform, not affiliated with or endorsed by Anthropic, offering CCDV-F preparation through adaptive study sessions, practice exams scored on the same 100-to-1000 scale with a 720 passing threshold, and Archie, a Socratic tutor that guides you through concepts with graduated hints rather than giving answers directly.

The adaptive engine uses Bayesian Knowledge Tracing with a 0.90 mastery threshold, routing you toward weaker domains rather than letting you over-practise material you already know. Given that Domain 2 alone accounts for a third of the exam, early routing toward Applications and Integration is where most candidates gain the most ground fastest.

Note that the CCDV-F concept library at /concepts is not yet live. What is available today: adaptive study sessions, Archie tutoring, and practice exams for CCDV-F, all scored and calibrated to the real exam's 720 passing bar.

What is the exam-day checklist for CCDV-F?

Before sitting the exam, work through the following:

1. Register via Pearson VUE and pay the $125 attempt fee.
2. Choose online-proctored delivery or a test centre.
3. Confirm the 120-minute window and 53-item count before you begin.
4. Review your weakest domains using the percent-correct breakdown from any practice exam you have completed.
5. Note that the credential is valid for 12 months from the award date and plan your renewal accordingly.

The raw-to-scaled score conversion is not published by Anthropic. Do not target a specific raw question count as a pass mark. Target 720 on the scaled report.