Writing guidelines skill: review docs against Vercel's writing handbook

Source: vercel-labs/agent-skills · Maintainer: Vercel · License: MIT per the repo README (no LICENSE file in the repo) · Checked: 24 October 2026

What is the writing-guidelines skill?

The writing-guidelines skill makes your coding agent review documentation and other prose against Vercel's writing handbook, then list every problem as a file:line finding: passive voice, banned words like "easy," missing code language tags, em dashes, vague claims and more. It is for developers and docs writers who want a consistent style review before a docs PR merges.

It sits in vercel-labs/agent-skills, Vercel's skills collection, next to the web design guidelines skill, and works the same way. The skill's description reads: "Review docs/prose for Writing Guidelines compliance. Use when asked to "review my docs", "check writing style", "audit prose", "review docs voice and tone", or "check this page against the writing handbook"."

The repo README says it "Audits your pages for 80+ rules covering voice, structure, content types, code samples, typography, and AI workflow." As of 24 October 2026, its skills.sh page listed about 105,100 installs (snapshot, 24 October 2026).

What does it do?

The folder holds a single SKILL.md of under 200 words. It contains no rules of its own. Instead it tells the agent to:

1. Fetch the latest rules from command.md in the separate vercel-labs/writing-guidelines repo, using WebFetch, before each review.
2. Read the files you named, or ask which files to review if you named none.
3. Check them against every rule in the fetched file.
4. Output findings in the terse file:line format the fetched file defines.

The rules live in command.md, which groups them by topic. Some of the categories, as of 24 October 2026:

The output is grouped by file, one line per issue, such as a missing language tag or a banned word with its line number. Files with no issues get a pass mark. The rules end with: "State issue + location. Skip explanation unless fix is non-obvious. No preamble."

How do you install it?

The skills.sh page gives this command for the skills CLI:

npx skills add https://github.com/vercel-labs/agent-skills --skill writing-guidelines

The repo README's command installs from the whole collection and lets you pick skills:

npx skills add vercel-labs/agent-skills

Add -g to install for every project. The folder lands in .claude/skills/writing-guidelines/ in your project, or ~/.claude/skills/writing-guidelines/ with -g, the standard Claude Code skill locations. Codex and Cursor read the same SKILL.md format; see Codex skills and Cursor skills.

On claude.ai, you can zip the folder and upload it under Customize > Skills, following the Claude help article on skills. The skill needs to fetch a URL, so check that web access works in that environment.

How do you use it?

Name the files or a glob. The skill is model-invoked, so phrases from its description, like "review my docs," can load it, or you can type /writing-guidelines. Example prompts (suggestions, not quotes from the repo):

• /writing-guidelines docs/getting-started.mdx
• Review my docs in content/docs/**/*.mdx against the writing handbook.
• Check writing style on README.md and CONTRIBUTING.md.
• Audit the prose in this PR's changed Markdown files and list file:line findings.
• Review docs voice and tone on the troubleshooting page, then fix every finding.

Requirements: the agent must be able to fetch web pages, since the rules come from GitHub at run time. No packages or keys are needed.

Tips and limits

• The rules change under you. Each run fetches command.md fresh, so two reviews months apart can apply different rules. The writing-guidelines repo was last updated on 1 June 2026. For repeatable results, copy command.md into your own skill and pin it.
• Some rules are Vercel house style. The guidelines README splits out "Vercel-specific" preferences, such as the ACME demo account and preferred model strings in examples. Some general rules, like meta.contentType and the <Steps/> component, also assume an MDX docs setup like Vercel's. Tell the agent which rules to skip.
• It reports; it doesn't fix. Ask for the fixes as a follow-up, and review them yourself.
• Security audit. As of 24 October 2026, skills.sh showed Pass from Gen Agent Trust Hub and Socket, and Warn from Snyk. Snyk's audit of 12 June 2026 rated it medium risk, with two findings: "Third-party content exposure detected (indirect prompt injection risk)" and "Unverifiable external dependency detected (runtime URL that controls agent)." Both stem from the runtime fetch of command.md. Pinning a local copy removes that dependency.
• License. The agent-skills README ends with "MIT", but as of 24 October 2026 the repo has no LICENSE file and GitHub reports no license for it. The vercel-labs/writing-guidelines repo, which holds the actual rules, has an MIT LICENSE file.

Save it in your Kipt library

In Kipt, open Skills and paste the skill's GitHub folder link into the import box. Kipt reads the SKILL.md; for a repo with many skills, like this one, it shows a list to pick from. You can also drop the folder or a .zip, for example one where you've added a pinned copy of command.md. Your skills live in your own private GitHub repo and install into Claude Code or Codex with one command:

npx skills add https://github.com/<you>/<repo>.git --all -g

Kipt is not affiliated with Vercel.

Similar skills

• Web design guidelines skill: the same fetch-and-review pattern for UI code.
• Humanizer skill: rewrites text to remove signs of AI writing.
• Doc co-authoring skill: Anthropic's workflow for drafting documents with Claude.
• Internal comms skill: Anthropic's skill for status updates, newsletters and FAQs.