[DOCS] Skills reference still implies `/verify` and `/code-review` can run automatically

Documentation Type

Incorrect/outdated documentation

Documentation Location

https://code.claude.com/docs/en/skills

Section/Topic

The introductory invocation guidance in Extend Claude with skills, the Bundled skills section, and the corresponding generic [Skill] definition in the commands reference.

Current Documentation

The skills page currently says:

Claude uses skills when relevant, or you can invoke one directly with /skill-name.

It also lists /code-review as a bundled skill and says:

You invoke them the same way as any other skill, by typing / followed by the skill name.

The commands reference describes a bundled skill this way:

Skill: a bundled skill. It works like skills you write yourself: a prompt handed to Claude, which Claude can also invoke automatically when relevant.

The same reference lists /verify and /code-review as bundled skills, but does not identify them as explicit-only commands.

What's Wrong or Missing?

Claude Code 2.1.215 changed the invocation behavior of two bundled skills:

Claude no longer runs the /verify and /code-review skills on its own; invoke them with /verify or /code-review when you want them

The current wording is now misleading for both commands. The skills page says Claude uses skills when relevant, and the commands reference explicitly says that a bundled skill can be invoked automatically when relevant. A reader can therefore expect Claude to select /verify or /code-review from the task context, even though 2.1.215 requires the user to invoke either skill explicitly.

This is one shared invocation-contract problem across the two pages: the general automatic-invocation description does not identify the /verify and /code-review exception. The page already teaches direct slash invocation, but it does not say that direct invocation is now required for these two bundled skills.

Suggested Improvement

Add an explicit invocation note to the Bundled skills section of the skills page, near the list that includes /verify and /code-review. For example:

Explicit invocation: As of Claude Code v2.1.215, Claude does not invoke /verify or /code-review automatically based on the task. Run /verify or /code-review yourself when you want to verify a change or review a diff.

Then narrow the generic definition in the commands reference so it does not apply automatic invocation to these two skills. For example:

[Skill]: a bundled skill. It works like skills you write yourself: a prompt handed to Claude. Most skills can be invoked automatically when relevant, but /verify and /code-review require an explicit slash-command invocation.

Keep the existing command rows and purpose descriptions, but add the same explicit-invocation note or a link back to the skills page so users who start from the commands table see the changed behavior. If other bundled skills retain automatic invocation, state that distinction rather than implying that all bundled skills share one invocation policy.

Impact

Medium - Makes feature difficult to understand

Additional Context

Affected Pages:

| Page | Line(s) | Context |
|------|---------|---------|
| https://code.claude.com/docs/en/skills | 9, 14-16, 21-27 | General skill invocation guidance and the bundled-skills list containing /code-review |
| https://code.claude.com/docs/en/commands | 35-40, 63, 148 | Generic [Skill] behavior and the /code-review and /verify command entries |

Total scope: 2 Claude Code documentation pages should describe the same explicit-invocation rule.

The release behavior is documented in the Claude Code changelog at https://code.claude.com/docs/en/changelog under version 2.1.215. The Platform Agent Skills documentation at https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview covers API and agent skill attachment rather than these Claude Code slash commands, so the correction should remain scoped to the Claude Code references above.