Quiz 1: Understanding Skills and the Specification
Test your understanding of skills and the Agent Skills Specification. Answer the following questions to check your knowledge before moving forward.
Question 1: What is a skill?
选项
- A. A skill is a long prompt you paste into every conversation
- B. A skill is a reusable package of knowledge (instructions, scripts, and references) that makes agents expert at a specific task
- C. A skill is a feature only available in Claude Code
- D. A skill is a Python package you install with pip
答案解析
- A — Not quite. While skills contain similar information to detailed prompts, they're structured and reusable. The key difference is that skills work across projects and teams automatically.
- B(正确答案) — Correct! Skills are portable, structured packages that include instructions, helper scripts, documentation, and metadata—all designed to make agents reliable at completing specific tasks.
- C — No, skills are supported by many agents including Claude Code, Codex, OpenCode, and others. The Agent Skills Specification is an open standard.
- D — No, skills are not Python packages. However, skills can contain Python scripts in their scripts/ directory, and they can live in normal project or team directories.
Question 2: What does the Agent Skills Specification define?
选项
- A. The best practices for writing Python code in helper scripts
- B. Requirements for one hosting or distribution platform only
- C. Guidelines for writing instruction text in skills
- D. A standard format for packaging agent knowledge, including directory structure, metadata fields, and how agents discover and load skills
答案解析
- A — Not quite. While the spec does cover script storage, it doesn't define Python best practices. The spec focuses on the overall skill structure and metadata format.
- B — No, the Agent Skills Specification is agent-agnostic and applies across all platforms (Claude Code, Codex, OpenCode, etc.). It is not tied to one distribution method.
- C — The spec covers some guidance, but it's primarily about the overall structure (directories, metadata, file names) rather than the instruction content itself.
- D(正确答案) — Correct! The Agent Skills Specification (at agentskills.io) defines the portable format that allows skills to work across different agents and platforms.
Question 3: What are the required frontmatter fields in a SKILL.md file?
选项
- A. name, description, version, and author
- B. Any metadata the creator wants to include
- C. name and description
- D. There is no required frontmatter—skills can be just Markdown files
答案解析
- A — While version and author are useful metadata, they're optional. Only 'name' and 'description' are required by the specification.
- B — Not quite. The spec defines specific required fields (name and description) to ensure compatibility and discoverability. Without these, agents can't properly identify and load the skill.
- C(正确答案) — Correct! The Agent Skills Specification requires two fields: 'name' (the identifier for the skill) and 'description' (what the skill does and when to use it). Other fields like license, compatibility, and metadata are optional.
- D — No, frontmatter is essential. It provides metadata that agents use to discover and load skills automatically. Without it, skills won't be recognized by compliant agents.
Question 4: How do agents discover and load skills?
选项
- A. Agents automatically load all available skills into context at startup
- B. Agents read skill metadata to decide if the skill applies, then load the full SKILL.md only when needed—this is called 'progressive disclosure'
- C. Users must manually activate skills by typing a command like '/skill dataset-publishing'
- D. Skills are only available if installed via npm or pip
答案解析
- A — No, that would be inefficient. Agents use progressive disclosure: they load only skill metadata initially, then activate full skills when the task matches.
- B(正确答案) — Correct! The discovery process is progressive: agents first check metadata (~100 tokens), then load full skill content only when the task matches. This keeps agent startup fast.
- C — While some agents support manual activation, the primary design is automatic discovery. Agents analyze the user's request and load relevant skills on their own.
- D — No, skills can be discovered from multiple sources: local directories, shared repositories, community registries, or custom paths. The Agent Skills Specification supports many discovery mechanisms.
Question 5: How many agents support the Agent Skills Specification?
选项
- A. Only Claude Code and Codex
- B. Only agents developed by Anthropic
- C. Only agents from a single vendor or registry
- D. Many agents, including Claude Code, Codex, OpenCode, and others
答案解析
- A — No, while Claude Code and Codex are major supporters, the spec is adopted by many agents including OpenCode and other commercial tools.
- B — No, the spec is open and platform-agnostic. It was originally created by Anthropic but is now maintained as an open standard and adopted by many organizations.
- C — No, skills can be used with agents available anywhere. Agents can load skills from local directories, team repositories, or other registries as long as they support the spec.
- D(正确答案) — Correct! The Agent Skills Specification is an open standard adopted by many agents. This means a single skill works across multiple platforms.
Scoring
Count your correct answers:
- 5/5: Excellent! You have a solid understanding of skills and the specification. Move on to the next lesson.
- 4/5: Good understanding. Review the questions you missed and reread those sections.
- 3/5 or less: Consider reviewing the "What Are Skills" and "Skill Format" lessons before continuing.
Ready to learn how to use skills with different agents? Let's move on to the next lesson.