What does this skill do, and when should you use it?
authoring-docs is a skill inside the e2e repository that governs writing for its documentation site (docs/**/*.mdx, excluding docs/reference/). It prescribes a page skeleton, icon rules, a component table for folding Mintlify content, and tone rules based on 80% ASD-STE100 Simplified Technical English. Its central mandate is a mandatory shortening pass before every commit: measure line and word counts against sidebar siblings, cut repetition, and fold what only some readers need. It also ships a ten-step process for rewriting legacy pages and a pre-commit checklist.
- Reads and analyzes MDX guide pages under docs/, measuring line counts and rough prose word counts with wc/awk and comparing against sidebar siblings
- Executes cuts in a fixed order: content the reader doesn't act on, statements said twice, then folding content only some readers need
- Produces pages on a fixed skeleton: lede, Before you start, task-named headings, a Next CardGroup of 2-4 cards
- Chooses folding components from a table: Tabs, CodeGroup, Steps, Accordion, ShowMore, Note, Warning
- Validates page icons: black silhouette SVGs under docs/images/icons/ that match sidebar neighbors in light and dark mode
- Verifies every fact against src/, runs the pre-commit checklist, and runs pnpm docs:check
- Writing a new integration guide page for the e2e docs site and wanting to follow the proven shape of quickstart.mdx and bug-bash.mdx
- Inheriting a long legacy MDX page that needs a structural rewrite rather than a copy edit
- A non-docs PR (like a flag rename) touching three pages, where cuts should apply only to the affected sections
- A user asks 'how do I shorten this page' and you want to show a plan first, with per-item line savings grouped by cut kind
- Picking a Font Awesome page icon or producing a brand silhouette icon that works in the sidebar
- e2e reference pages — the skill explicitly excludes docs/reference/, which are lookup tables keeping their own shape
- Content under skills/e2e/ — the skill declares agents read it start to finish and interactive components don't render there
- Documentation projects outside the e2e repo — path conventions (docs/images/icons/, docs/docs.) and scripts (pnpm docs:check) are repo-specific and would need adaptation
How do you install this skill?
- Publisher is unverified; treat identity as unknown and confirm repository provenance yourself.
- This is a static source-only review: no commands or validation scripts were executed, so all scores are low confidence.
- The skill depends on repo-internal conventions (Mintlify components, docs:check, the unslop skill, docs.); portability outside this repository is unknown.
- Referenced example pages (bug-bash.mdx, quickstart.mdx) were not part of the evidence, so their related claims were not cross-checked.
- The docs site and npm ecosystem are overseas; mainland-China users may face access limits for those dependencies.
- Shell / CLI
- Local filesystem
pnpmMintlify docs toolchain (docs:check, docs:dev scripts)
The source provides no installation commands for this skill. The README only states the repository bundles 10 skills, with this one at .dev/skills/authoring-docs/SKILL.md, but documents no install or load commands (such as a plugin add or copy step), so no steps are invented here.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/tester-army/e2e.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/.dev/skills/authoring-docs" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Write a new e2e integration guide at docs/integrations/xxx.mdx following the skill's page skeleton
- docs/ci/cache.mdx is twice as long as its sidebar neighbors — run the shortening check and show a cut plan with per-item line savings
- This PR renames a flag and updates three doc pages — apply cuts only to the touched sections and list the rest as follow-up
- Rewrite the legacy page docs/migrate/legacy.mdx keeping every fact, following the ten-step process
The skill is triggered automatically by its frontmatter description whenever the task involves writing, editing, rewriting, or shortening a docs guide page (docs/**/*.mdx outside docs/reference/); metadata marks it internal. The flow: write the target outline first; run the shortening pass on every page (wc -l for lines, awk filtering code fences then wc -w for rough prose words); sort cuts into the three kinds and work them in order; when asked how to shorten a page, show the plan grouped by kind with rough line savings before editing; then work through the pre-commit checklist, run pnpm docs:check, and preview with pnpm docs:dev the way a skimmer would.
What are this skill's strengths and limitations?
- Extremely concrete and executable: measurement commands, a component-selection table, sentence-length limits, and a banned-words table rather than vague style advice
- Built-in drift protection: every new or merged sentence must be checked against src/, and every cut fact must be named with its reason in the PR body
- Clear scope boundaries (not reference, not skills/e2e/) and an explicit precedence rule over the unslop skill on docs pages
- A complete ten-step legacy-page rewrite process including anchor-link stability checks, covering real maintenance scenarios
- Tightly coupled to the e2e repo's file paths, Mintlify components, and pnpm scripts; significant adaptation needed for other projects
- No installation commands or adoption evidence in the source, so real-world triggering can't be verified
- The repository is in active pre-1.0 development, so referenced pages (like bug-bash.mdx) may change
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| e2e Docs Authoring Skill this page | 54 · Use with care | ★ 8.7k | 1d ago | Apache-2.0 |
| Khazix WeChat Long-Form Writer Skill | 49 · Use with care | ★ 21k | 10d ago | MIT |
| Shuorenhua: Chinese AI-Tone Cleanup Skill | 65 · Recommended | ★ 2k | 11d ago | MIT |
| No AI Slop | 59 · Recommended | ★ 12k | 1mo ago | MIT |
| Beautiful Prose Style Contract | 50 · Use with care | ★ 57 | 9mo ago | — |
The skill body notes its relationship to the unslop skill: unslop applies to every sentence, while this skill adds docs structure and tone on top. Where they disagree, this skill wins on docs pages — though the STE sentence and paragraph limits still apply and no first person is allowed.
How did FollowSkills review this skill?
The skill itself is only a docs-authoring guide: no commands executed, no permissions requested, no external side effects. At repo level, the trust model is clear (SECURITY.md separates code-trust from untrusted model input, telemetry is opt-out with full field disclosure, no crash reporting or update checks). Deductions: the skill itself does not declare data flows (e.g. requires cross-checking src/ and referencing an external docs site), publisher identity is unverified, and rollback/confirmation relies on the PR process rather than in-skill statements.
High internal consistency: clear scope (guides vs reference vs skills/e2e), a reproducible measurement procedure (wc/awk), and a complete pre-commit checklist. Deductions: static review cannot execute any validation; several checks (pnpm docs:check, typecheck, unslop) rely on scripts not shown in evidence; failure-feedback quality on abnormal input (missing pages, failed commands) is unverifiable, so above 10 is unreachable per anchors.
Trigger description is precise: explicit scenarios (docs/**/*.mdx excluding reference/), explicit exclusions (reference pages, skills/e2e), model pages and a skeleton given. Deductions: boundaries partially depend on repo-internal conventions (Mintlify components, docs. redirects, the unslop skill); portability outside this repo is undeclared; Chinese support not addressed; the docs site and npm ecosystem are overseas but the core function (writing guidance) does not depend on online services, so only a small deduction.
Well-structured with progressive disclosure (principles, then process, then checklist), examples and before/after comparisons. Deductions: the skill file itself has no version, changelog, or known-limitations section; maintenance responsibility is implicit in repo governance rather than stated in the skill; name/description match capability and license (Apache-2.0) is clear.
The goal is clear (write/edit/shorten docs pages); the output is a directly followable editing process and checklist with marginal value over unguided rewriting. Deductions: static review cannot verify actual rewriting outcomes; evidence of effectiveness is only self-declared process plus references to example pages (bug-bash.mdx, quickstart.mdx) not included in evidence, so above 7 is unreachable.
Key claims are partially traceable: concrete file paths, measurement commands, check scripts (check-docs-examples.ts, docs:check), existing CI workflows (benchmark, agent) and committed test suites, and detailed security/telemetry claims. Deductions: the skill's own key paths (writing outcomes, shortening results) lack executable third-party verification; the referenced example pages are not in evidence; static review caps this at 5.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →