Sources

No — Required: No. "Display name shown in skill listings. Defaults to the directory name." A skill with no name still registers; the command comes from the directory.

Claude Code skill frontmatter: is `name` required?

Anthropic — Extend Claude with skills, Frontmatter reference. Read 2 August 2026. Not yet re-checked automatically, so this one is confirmed by reading it.

No — Required: Recommended. "If omitted, uses the first paragraph of markdown content."

Claude Code skill frontmatter: is `description` required?

Anthropic — Extend Claude with skills, Frontmatter reference. Read 2 August 2026. Not yet re-checked automatically, so this one is confirmed by reading it.

1,536 characters, applied to `description` and `when_to_use` COMBINED — "truncated at 1,536 characters in the skill listing to reduce context usage". As of the 2026-08-31 reading the cap is the DEFAULT rather than a fixed limit: "The cap is configurable with `skillListingMaxDescChars`", and the surrounding listing budget with `skillListingBudgetFraction` / SLASH_COMMAND_TOOL_CHAR_BUDGET.

Documented character cap on a skill's listing text

Anthropic — Extend Claude with skills, Frontmatter reference. Read 31 August 2026. Not yet re-checked automatically, so this one is confirmed by reading it.

Yes, both. `name` Required: Yes ("Unique identifier using lowercase letters and hyphens"); `description` Required: Yes ("When Claude should delegate to this subagent").

Claude Code subagent frontmatter: are `name` and `description` required?

Anthropic — Create custom subagents, frontmatter reference. Read 2 August 2026. Not yet re-checked automatically, so this one is confirmed by reading it.

Markdown files in `.claude/agents/` (project) or `~/.claude/agents/` (user), scanned RECURSIVELY — subfolders such as `agents/review/` are supported and the subdirectory path does not affect identity or invocation. NOT true inside a plugin: plugin `agents/` directories are scanned recursively too, but there the subfolder DOES become part of the identifier — `agents/review/security.md` in plugin `my-plugin` registers as `my-plugin:review:security`.

Where a Claude Code subagent is defined

Anthropic — Create custom subagents. Read 2 August 2026. Not yet re-checked automatically, so this one is confirmed by reading it.

`SKILL.md` (required) in `~/.claude/skills/<skill-name>/` (personal) or `.claude/skills/<skill-name>/` (project).

Where a Claude Code skill is defined

Anthropic — Extend Claude with skills. Read 2 August 2026. Not yet re-checked automatically, so this one is confirmed by reading it.

`.claude-plugin/marketplace.json` in the repository root. Users add it with `/plugin marketplace add owner/repo`.

The file a repository must publish to act as a Claude Code plugin marketplace

Anthropic — Create and distribute a plugin marketplace. Read 2 August 2026. Not yet re-checked automatically, so this one is confirmed by reading it.

Skills, agents, hooks, MCP servers or LSP servers. Custom commands have been MERGED INTO SKILLS — `.claude/commands/deploy.md` and `.claude/skills/deploy/SKILL.md` both create `/deploy`.

What a Claude Code plugin can bundle

Anthropic — Create and distribute a plugin marketplace. Read 2 August 2026. Not yet re-checked automatically, so this one is confirmed by reading it.

No — "Claude Code skills follow the Agent Skills open standard, which works across multiple AI tools", with Claude Code extending it (invocation control, subagent execution, dynamic context injection).

Whether Claude Code skills are a proprietary format

Anthropic — Extend Claude with skills. Read 2 August 2026. Not yet re-checked automatically, so this one is confirmed by reading it.

`strict` controls "whether `plugin.json` is the authority for component definitions" and defaults to true. Under `true` the two sources merge: the marketplace entry "can supplement it with additional components, and both sources are merged". Under `false` the marketplace entry "is the entire definition", and "if the plugin also has a `plugin.json` that declares components, that's a conflict and the plugin fails to load".

When a Claude Code plugin fails to load because of the marketplace entry's `strict` field

Anthropic — Create and distribute a plugin marketplace, Strict mode. Read 9 August 2026. Not yet re-checked automatically, so this one is confirmed by reading it.

Yes. Names "can't contain `:`, which is reserved for plugin-scoped identifiers such as `my-plugin:reviewer`. Claude Code doesn't load a file whose name contains one and logs an error to the debug log. Before v2.1.218, such names were accepted." The skill frontmatter reference documents no reserved character on `name`, so this is subagent-only.

Whether a colon in a Claude Code subagent `name` stops it loading

Claude Code docs — Create custom subagents, frontmatter reference. Read 31 August 2026. Not yet re-checked automatically, so this one is confirmed by reading it.

A fixed bundled set. "Claude Code includes a set of bundled skills, such as /doctor, /code-review, /batch, /debug, /loop, and /claude-api." They are prompt-based rather than compiled: "Bundled skills are prompt-based: they give Claude detailed instructions and let it orchestrate the work using its tools. Most built-in commands instead execute fixed logic directly." "Most bundled skills are available in every session"; a few depend on a feature, as /workflow-authoring depends on dynamic workflows. Three more, /run, /verify and /run-skill-generator, launch and verify your app.

Which skills ship with Claude Code, and what kind of thing they are

Anthropic — Extend Claude with skills, Bundled skills. Read 6 September 2026. A job re-fetches this page and looks for the wording the figure came from; it last found it 23 September 2026.

Yes for the name, no for the alias. The precedence table row for "Any of those locations and a bundled skill" reads: "Your skill replaces the bundled command, but not its aliases. A project code-review skill replaces /code-review, and the bundled alias /review never runs your skill."

Does an installed skill override a bundled skill of the same name?

Anthropic, Extend Claude with skills, Resolve skills that share a name. Read 18 September 2026. A job re-fetches this page and looks for the wording the figure came from; it last found it 23 September 2026.

No. The precedence table row for "A plugin skill and a skill at any of the locations above" reads: "Both load, because plugin skills are namespaced as /plugin-name:skill-name". my-plugin/skills/deploy/SKILL.md becomes /my-plugin:deploy and loads alongside a deploy skill in a project's .claude/skills/.

Can a plugin skill collide with a skill at another level?

Anthropic, Extend Claude with skills, Resolve skills that share a name. Read 18 September 2026. A job re-fetches this page and looks for the wording the figure came from; it last found it 23 September 2026.

"Enterprise over personal, and personal over project. With deploy in both ~/.claude/skills/ and the project's .claude/skills/, /deploy runs the personal one." Against a skill synced from a claude.ai account the local one wins, listed as "The other skill or command", and the synced skill still runs as /anthropic-skills:<name>. Where a .claude/commands/ file and a .claude/skills/ skill share a name, the table answers "The skill".

Precedence when skills at different levels share a name

Anthropic, Extend Claude with skills, Resolve skills that share a name. Read 18 September 2026. A job re-fetches this page and looks for the wording the figure came from; it last found it 23 September 2026.

Yes, and /doctor survives it. "To turn bundled skills off, use the disableBundledSkills setting. The /doctor setup checkup stays typable when disableBundledSkills is on, in Claude Code v2.1.205 and later. To hide it, set the DISABLE_DOCTOR_COMMAND environment variable or a skillOverrides entry of "doctor": "off"." Before v2.1.205, /doctor was a built-in command rather than a bundled skill.

Can bundled skills be turned off?

Anthropic, Extend Claude with skills, Bundled skills. Read 18 September 2026. A job re-fetches this page and looks for the wording the figure came from; it last found it 23 September 2026.

Standalone first. The documented table gives standalone .claude/ as best for "Personal workflows, project-specific customizations, quick experiments" and plugins as best for "Sharing with teammates, distributing to community, versioned releases, reusable across projects", followed by: "Start with standalone configuration in .claude/ for quick iteration, then convert to a plugin when you're ready to share."

Standalone .claude/ configuration or a plugin: which does Anthropic document first?

Anthropic — Create plugins, When to use plugins vs standalone configuration. Read 6 September 2026. A job re-fetches this page and looks for the wording the figure came from; it last found it 23 September 2026.