Agent skill

NeMo Curator Docs Maintenance

by NVIDIA-NeMo in NVIDIA-NeMo/Curator

Adds, updates, moves and removes pages on the NeMo Curator Fern documentation site, keeping navigation entries, links and redirects in step.

Apache-2.0Auto-check passedDevelopment

Install NeMo Curator Docs Maintenance

skills CLI
$ npx skills add NVIDIA-NeMo/Curator --skill nemo-curator-docs -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install NVIDIA-NeMo/Curator nemo-curator-docs --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/NVIDIA-NeMo/Curator.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/nemo-curator-docs .claude/skills/nemo-curator-docs && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
nemo-curator-docs
GitHub stars
1.8k
Token cost
~4.3k tokens
SKILL.md length
1,506 words
Files
1
Skills in repo
2
Repo updated
First seen
Licence
Apache-2.0

At a glance

Adds, updates, moves and removes pages on the NeMo Curator Fern documentation site, keeping navigation entries, links and redirects in step.

  • Works in 4 steps: Gather: page title, target section,… → Create fern/versions/v26.04/pages//.mdx → Add a nav entry in… → …
  • Adding a new documentation page to the NeMo Curator Fern site
  • SKILL.md covers Scope Rule, Layout at a Glance, Operations and Content Guidelines, plus 11 more sections
  • Calls git, npm and python; reaches github.com; needs FERN_TOKEN

What it does

All documentation work happens under `fern/`; the old `docs/` directory is deprecated, and even release notes and migration guides go in the new tree. New pages default to the current version `v26.04`, which `latest` points to through a symlinked config, unless you name another version. A page is a kebab-case `.mdx` file with frontmatter, plus an entry in `fern/versions/v26.04.yml`.

Updating covers content edits, title changes in both frontmatter and navigation, section moves with `git mv` and link fixes, and slug changes with a redirect in `fern/docs.yml`. Removing a page means finding incoming links with grep, running `git rm`, deleting its navigation block (and the section if it empties), fixing links and adding a redirect when the URL was public. A further section deals with back-porting to an older version.

When your agent uses it

  • Adding a new documentation page to the NeMo Curator Fern site
  • Renaming or moving a page without breaking existing links
  • Removing an outdated page and cleaning up navigation and redirects

Example prompts

  • “Add a page called Text Deduplication to the data processing section of the Curator docs.”
  • “Move the quickstart page into the getting-started section and add the redirect.”
  • “Remove the legacy install page and fix every link that points to it.”

Requirements

  • A checkout of the NeMo Curator repository with its `fern/` docs tree

Workflow steps

4 steps, taken from the first numbered list in SKILL.md.

  1. Gather: page title, target section, filename (kebab-case .mdx), subdirectory under fern/versions/v26.04/pages/.
  2. Create fern/versions/v26.04/pages//.mdx
  3. Add a nav entry in fern/versions/v26.04.yml under the correct section
  4. If this also applies to latest, no action needed — latest.yml is a symlink to v26.04.yml.

What it can do on your machine

Read from SKILL.md and the folder at commit 77ec4a7. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • git
    • npm
    • python
    • python3

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Hosts in commands or code, which the agent is likely to contact:

    • github.com

    Also links to:

    • buildwithfern.com

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names these keys or tokens, usually read from environment variables:

    • FERN_TOKEN

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

NeMo Curator Docs Maintenance loads about 4.3k tokens when it runs. Until then it costs about 36 tokens; SKILL.md has 1,506 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~36
When it runs · the whole SKILL.md, loaded when a task matches
~4.3k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from NVIDIA-NeMo/Curator at commit 77ec4a7, republished under its Apache-2.0 licence (© NVIDIA-NeMo). 1,506 words, ~4,301 tokens.

Download SKILL.mdSave it as .claude/skills/nemo-curator-docs/SKILL.md (or your agent's skills folder).
name
nemo-curator-docs
description
Maintain the NeMo Curator Fern docs site — add, update, move, or remove pages under fern/. Use for any documentation changes.

NeMo Curator Docs Maintenance

Unified skill for adding, updating, moving, and removing pages on the NeMo Curator Fern documentation site.

Scope Rule

ALL docs edits happen under fern/. The legacy docs/ directory is deprecated — do not add or move content into it. Release notes, migration guides, and every new page belong under fern/.

Layout at a Glance

fern/
├── fern.config.json          # Minimal Fern config (org + CLI version)
├── docs.yml                  # Site config: versions, tabs, redirects, libraries
├── versions/
│   ├── latest.yml            # Symlink → v26.04.yml (do not edit directly)
│   ├── v26.04.yml            # Nav tree for current train
│   ├── v26.04/pages/         # MDX content for current train
│   ├── v25.09.yml
│   └── v25.09/pages/
├── components/               # Custom TSX components (footer, etc.)
├── assets/                   # Images, SVGs, favicon
├── substitute_variables.py   # CI: resolves {{ variables }} in MDX
└── AUTODOCS_GUIDE.md         # Library reference generation guide

Current train: v26.04. Default all new pages there unless the user specifies a version.

File system                              Published URL
───────────────────────────────────────  ────────────────────────────────────────
fern/versions/v26.04/pages/              docs.nvidia.com/nemo/curator/latest/
  └─ get-started/text.mdx                  └─ get-started/text
fern/versions/v26.04.yml ── nav for ──┐  docs.nvidia.com/nemo/curator/v26.04/
fern/versions/latest.yml ─ symlink ───┘    └─ get-started/text
fern/versions/v25.09/pages/              docs.nvidia.com/nemo/curator/v25.09/
  └─ get-started/text.mdx                  └─ get-started/text

Operations

Add a Page
  1. Gather: page title, target section, filename (kebab-case .mdx), subdirectory under fern/versions/v26.04/pages/.
  2. Create fern/versions/v26.04/pages/<subdirectory>/<filename>.mdx:
mdx
---
description: "One-line SEO description"
categories: ["<category>"]
tags: ["<tag-1>", "<tag-2>"]
personas: ["<persona>"]
difficulty: "beginner"      # beginner | intermediate | advanced
content_type: "tutorial"     # tutorial | how-to | reference | concept | index
modality: "text-only"        # text-only | image-only | video-only | audio-only | universal
---

# <Page Title>

<content>
  1. Add a nav entry in fern/versions/v26.04.yml under the correct section:
yaml
- page: <Page Title>
  path: ./v26.04/pages/<subdirectory>/<filename>.mdx
  slug: <filename>
  1. If this also applies to latest, no action needed — latest.yml is a symlink to v26.04.yml.
Update a Page
  1. Locate by path, title, or keyword (grep -rn in fern/versions/v26.04/pages/).
  2. Content only — edit the MDX directly.
  3. Title change — update the frontmatter and the - page: name in fern/versions/v26.04.yml.
  4. Section move — git mv the file, update its path: in the nav, and fix all incoming links.
  5. Slug change — update slug: in the nav and add a redirect in fern/docs.yml so old URLs keep working.
Remove a Page
  1. Find incoming links: grep -r "<filename>" fern/versions/v26.04/pages/ --include="*.mdx".
  2. git rm fern/versions/v26.04/pages/<subdirectory>/<filename>.mdx.
  3. Remove the - page: block from fern/versions/v26.04.yml. If it was the last page in a section, remove the - section: block.
  4. Fix or remove all incoming links found in step 1.
  5. Add a redirect in fern/docs.yml if the URL was public.
Back-port to an Older Version

Only when explicitly asked. Repeat the operation in the corresponding fern/versions/vXX.YY/ tree and vXX.YY.yml nav. MDX content often diverges between trains — do not blindly copy.

Worked Example: Adding a Page

Request: "Add a how-to for benchmarking text pipelines under Curate Text."

  1. Create fern/versions/v26.04/pages/curate-text/benchmarking.mdx:

    mdx
    ---
    description: "Benchmark text curation pipelines and interpret throughput and memory metrics"
    categories: ["how-to"]
    tags: ["text-curation", "benchmarking", "performance"]
    personas: ["mle-focused"]
    difficulty: "intermediate"
    content_type: "how-to"
    modality: "text-only"
    ---
    
    # Benchmark Text Pipelines
    
    <content>
  2. Add nav entry in fern/versions/v26.04.yml under the existing Curate Text section:

    yaml
    - page: Benchmark Text Pipelines
      path: ./v26.04/pages/curate-text/benchmarking.mdx
      slug: benchmarking
  3. cd fern && fern check then fern docs dev and verify the page renders at /curate-text/benchmarking.

Worked Example: Renaming a Slug (with Redirect)

Request: "Rename /curate-text/benchmarking to /curate-text/performance."

  1. Update slug: in fern/versions/v26.04.yml: slug: performance.

  2. (Optional) git mv the MDX file if you want the filename to match the slug.

  3. Add a redirect to fern/docs.yml so old links keep working:

    yaml
    redirects:
      - source: "/nemo/curator/latest/curate-text/benchmarking"
        destination: "/nemo/curator/latest/curate-text/performance"
      - source: "/nemo/curator/v26.04/curate-text/benchmarking"
        destination: "/nemo/curator/v26.04/curate-text/performance"
  4. grep -rn "/curate-text/benchmarking" fern/versions/v26.04/pages/ and update any incoming links.


Content Guidelines

NeMo Curator uses Fern-native MDX components directly (unlike Dynamo, which converts GitHub callouts in CI). Do not use > [!NOTE] syntax — it will not render.

PurposeComponent
Neutral aside<Note>...</Note>
Helpful tip<Tip>...</Tip>
Informational callout<Info>...</Info>
Warning<Warning>...</Warning>
Error / danger<Error>...</Error>
Card grid on index pages<Cards> with <Card title="..." href="..."> children

Images live in fern/assets/ (shared) or fern/versions/vXX.YY/pages/_images/ (version-scoped). Reference with root-relative paths.

Component examples:

mdx
<Tip>
If `uv` is not installed, see the [Installation Guide](/admin/installation).
</Tip>

<Warning>
GPU-accelerated dedup requires CUDA {{ recommended_cuda }} or later.
</Warning>

<Cards>
  <Card title="Text Curation" href="/get-started/text">
    Set up and run text curation workflows.
  </Card>
  <Card title="Image Curation" href="/get-started/image">
    Set up and run image curation workflows.
  </Card>
</Cards>

Frontmatter Fields

Required: description. Optional but strongly preferred: categories, tags, personas, difficulty, content_type, modality. Existing pages in the same section are the best reference for valid values.

title is taken from the - page: entry in the nav file; the MDX file itself uses an # H1 heading matching the page name.

Variable Substitution

Tokens like {{ product_name }}, {{ container_version }}, {{ current_release }}, {{ github_repo }}, {{ min_python_version }} are resolved by fern/substitute_variables.py at CI time. Use them instead of hard-coding versions or URLs. Canonical list in DEFAULT_VARIABLES at the top of that file.

Example in MDX:

mdx
Install {{ product_name }} {{ current_release }} from {{ github_repo }}.
Requires Python {{ min_python_version }}+ and CUDA {{ recommended_cuda }}.

After substitution at CI time:

Install NeMo Curator 25.09 from https://github.com/NVIDIA-NeMo/Curator.
Requires Python 3.10+ and CUDA 12.0+.

To preview substitution locally:

bash
python fern/substitute_variables.py versions/v26.04 --version 26.04 --dry-run

Validate

bash
cd fern
fern check                   # YAML + frontmatter validation
fern docs broken-links       # link check
fern docs dev                # localhost:3000 hot-reload preview

fern check must pass before commit. Broken-link check can be deferred but must pass in CI.

Commit & Preview

bash
git add fern/
git commit -s -m "docs: <add|update|remove> <page-title>"

DCO sign-off (-s) is required on every commit. CI enforces it. If you forget, amend with git commit --amend --no-edit -s and force-push the branch.

PRs that touch fern/** get an automatic Fern preview URL posted as a comment by .github/workflows/fern-docs-preview-comment.yml. No manual step needed.

                    ┌─ fern-docs-ci.yml         → fern check + autodocs
PR (touches fern/) ─┼─ fern-docs-preview.yml    → preview build
                    └─ fern-docs-preview-*.yml  → 🌿 preview URL comment

Merge to main      → NO publish. Site is unchanged.

Tag push (docs/v*) → publish-fern-docs.yml      → docs.nvidia.com/nemo/curator

Publishing to Production

Merging to main does NOT publish. Production only updates when a tag matching docs/v* is pushed (or the workflow is manually dispatched from the Actions tab). Do not push tags unless the user asks.

Tag must be docs/v<MAJOR>.<MINOR>.<PATCH> — the docs/v prefix is required by the workflow trigger and the semver suffix should match the docs release in CHANGELOG.md.

bash
# Correct — triggers publish
git tag docs/v1.1.0
git push origin docs/v1.1.0

git tag docs/v1.2.0-rc1     # pre-release suffix is fine, still matches docs/v*
git push origin docs/v1.2.0-rc1

# Wrong — these will NOT trigger publish
git tag v1.1.0              # missing docs/ prefix
git tag docs/1.1.0          # missing v
git tag docs-v1.1.0         # wrong separator

URL → version mapping after publish:

docs.nvidia.com/nemo/curator/latest/...   → symlink to current train (v26.04 today)
docs.nvidia.com/nemo/curator/v26.04/...   → 26.04 train
docs.nvidia.com/nemo/curator/v26.02/...   → 26.02 train
docs.nvidia.com/nemo/curator/v25.09/...   → 25.09 train

Version Ship Checklist (when cutting a new train)

When the user ships a new version (e.g. cutting v26.06 while v26.04 is current):

  1. Copy fern/versions/v26.04/pages/ → fern/versions/v26.06/pages/ and edit content.
  2. Copy fern/versions/v26.04.yml → fern/versions/v26.06.yml and update all ./v26.04/ path prefixes to ./v26.06/.
  3. Repoint the symlink: ln -sf v26.06.yml fern/versions/latest.yml.
  4. Update fern/docs.yml versions: list — add the new display-name, mark older trains stable.
  5. Add redirect rules in fern/docs.yml for /nemo/curator/26.06/:path* → /nemo/curator/v26.06/:path* (see existing patterns).
  6. Add */index.html redirect for the new version (e.g. /nemo/curator/v26.06/index.html → /nemo/curator/v26.06). The :path* rule does not match the empty-path case, so each version-root index.html needs its own explicit rule.
  7. Align display-name strings with CHANGELOG.md and nemo_curator/package_info.py.
Show full SKILL.md (724 more words)Show less

Holding a Version Back from Publish

A version is included in the published site only when it appears in the versions: block of fern/docs.yml. The MDX tree (fern/versions/vXX.YY/) and nav file (fern/versions/vXX.YY.yml) can sit in the repo unpublished — Fern doesn't auto-publish every YAML it finds.

To stage a version without publishing it (e.g. work-in-progress on v26.06 while v26.04 is current):

yaml
# fern/docs.yml
versions:
  - display-name: "Latest · v1.1.2 (26.04)"
    path: versions/latest.yml
    slug: latest
  - display-name: "26.04 · v1.1.2"
    path: versions/v26.04.yml
    slug: v26.04
  # v26.06 staged in repo but not listed here → not published

To pull an already-shipping version back (e.g. hold v26.04 while pushing fixes to older trains):

  1. Remove the v26.04 entry from versions: in fern/docs.yml.
  2. If latest should also stop serving 26.04 content, repoint the symlink: ln -sf v26.02.yml fern/versions/latest.yml. Otherwise leave latest alone — it will keep serving v26.04 content under /latest/ even with v26.04 removed (since latest.yml references the v26.04 pages directly).
  3. Tag and push docs/v* to publish.
  4. Restore the entry (and symlink) when ready.

This is a temporary maneuver — track the change so it gets reverted.

Audiences (alternative): Fern supports audiences: on versions plus separate instances: (e.g. staging vs production). This is heavier setup — only adopt if multi-instance publishing is genuinely needed. NeMo Curator does not currently configure instances. References: Fern versioning, Fern publishing.

Do not use hidden: true to hide a version from publish. Hidden versions are removed from navigation/search/indexing but remain accessible by direct URL — still effectively published.

Library Reference (Autodocs) and the Fern Cross-Ref Bug

fern/docs.yml declares one libraries: entry, and a script provides a second, temporary one (both write to fern/product-docs/nemo-curator/Full-Library-Reference/, gitignored):

  • nemo-curator (input.git, committed in docs.yml) — used by CI/publish/preview; requires fern login or FERN_TOKEN. Workflows run fern docs md generate --library nemo-curator.
  • nemo-curator-local (input.path: ../nemo_curator, injected temporarily by fern/scripts/generate-library-local.sh) — local dev without Fern auth; requires Docker. Run npm run generate:library:local from fern/. The entry must NOT be committed to docs.yml: fern docs dev rejects path-input libraries ("'path' input which is not yet supported") and renders a blank page, while --local generation only accepts path inputs.

For a full API reference locally (424 pages), use npm run generate:library. The local Docker parser is beta and currently produces fewer pages — fine for testing the no-auth flow, not a substitute for CI generation.

Known bug in the Fern Python library generator (filed upstream): the generator emits cross-references that miss the /nemo/curator site basepath (links use /nemo-curator/... instead of /nemo/curator/nemo-curator/...) and tacks on Sphinx-style #nemo_curator-… fragments that don't match any rendered anchor. Result: ~540 broken links across the generated API reference.

No in-repo workaround currently. A post-generation rewrite (walking the generated MDX, fixing the basepath, dropping stale fragments) is feasible but not yet wired up. Track the upstream Fern fix; revisit if it doesn't land soon.

fern/_fix_broken_links.py separately rewrites a long list of legacy URL patterns (/api/reference/api-reference/, old Sphinx slugs, etc.) on the committed MDX under fern/versions/v25.09/pages/ and fern/versions/v26.02/pages/. CI does not run it, so committed pages can drift. Re-run locally and commit the diff if you see drift:

bash
python3 fern/_fix_broken_links.py

Redirect Quirks

  • :path* does not match the empty-path case. /nemo/curator/:path*/index.html will not catch /nemo/curator/index.html — that needs its own explicit rule. Same for every version-root: /nemo/curator/{latest,vXX.YY}/index.html each need a dedicated entry. Pattern: define the explicit empty-path rules before the :path* rule.
  • Order matters. Fern processes redirects top-down, first match wins. Put more specific rules above catch-alls.
  • Version slugs in fern/docs.yml use the vXX.YY form (e.g. v26.04). Calendar-train forms (26.04) need redirects to the v-prefixed slug.

Debugging

SymptomFix
fern check YAML error2-space indent; - page: inside contents:; path: is relative to the version YAML file
Page 404 in previewslug: missing or duplicated in the same section; confirm in vXX.YY.yml
{{ variable }} shows literally on siteNot in DEFAULT_VARIABLES in substitute_variables.py — add it there
MDX parse errorReplace bare <https://...> with [text](https://...); escape < in prose with &lt; or backticks
Old Sphinx URL breaksAdd a redirects: entry in fern/docs.yml
Library reference missingRun npm run generate:library in fern/ (full reference; needs Fern auth). Or npm run generate:library:local without auth (Docker; beta, fewer pages). See fern/AUTODOCS_GUIDE.md.
Broken imagePath is relative to the MDX file; check fern/assets/ or pages/_images/ exists

Key References

FilePurpose
fern/docs.ymlSite config, versions, redirects, libraries
fern/versions/vXX.YY.ymlNavigation tree for a version
fern/versions/vXX.YY/pages/MDX content for a version
fern/versions/latest.ymlSymlink → current train's nav (do not edit)
fern/components/Custom TSX (footer, release banner)
fern/assets/Shared images, SVGs, favicon
fern/substitute_variables.pyVariable definitions + CI replacement
fern/AUTODOCS_GUIDE.mdGenerating library reference MDX from source
fern/README.mdFull docs architecture guide
.github/workflows/fern-docs-*.ymlCI: validation, preview, publish

© NVIDIA-NeMo, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .claude/skills/nemo-curator-docs of NVIDIA-NeMo/Curator.

Open the folder on GitHubat commit 77ec4a7

Compare with similar skills

NeMo Curator Docs Maintenance next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

NeMo Curator Docs Maintenance compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
NeMo Curator Docs Maintenance this skillNVIDIA-NeMo/Curator1.8k—~4.3kAutomated safety check: PassApache-2.0
Release Docs SyncUfoMiao/zcf6.1k—~4kAutomated safety check: PassMIT
Doc SyncJetBrains/ideavim10k2 repos~2.6kAutomated safety check: PassMIT
Mermaid Diagramsjjmartres/opencode1336 repos~1.9kAutomated safety check: PassMIT
Acquire Codebase Knowledgegithub/awesome-copilot40k1 repos~2.3kAutomated safety check: PassMIT
Releasejrswab/axe895—~1.4kAutomated safety check: PassApache-2.0

Similar skills

  • Compares code changes since the last Git tag with the multilingual docs and CLAUDE.md, then updates them or only reports mismatches with --check-only.

    6.1k GitHub stars~4k tokensUpdated 1 mo ago
    DevelopmentAuto-check passed
  • Doc Sync

    JetBrains/ideavim

    Official

    Keeps IdeaVim documentation in sync with code changes. An agent skill from JetBrains/ideavim.

    10k GitHub starsUsed in 2 repos~2.6k tokens
    DevelopmentAuto-check passed
  • Mermaid Diagrams

    jjmartres/opencode

    Helps an agent pick the right Mermaid diagram type and write the syntax for class, sequence, flow, ER, C4, state and other software diagrams.

    133 GitHub starsUsed in 6 repos~1.9k tokens
    DevelopmentAuto-check passed
  • Acquire Codebase Knowledge

    github/awesome-copilot

    Official

    Maps an unfamiliar codebase into seven evidence-backed documents in docs/codebase/, using a scan script and templates, for onboarding or architecture write-ups.

    40k GitHub starsUsed in 1 repo~2.3k tokens
    DevelopmentAuto-check passed
  • Release

    jrswab/axe

    Prepare code for release (version bumps, changelog, README updates) and create an annotated tag to trigger the GoReleaser workflow.

    895 GitHub stars~1.4k tokensUpdated 3 days ago
    DevelopmentAuto-check passed
  • Commit

    LennartHennigs/Button2

    Stage and commit current changes for Button2 — checks for needed CHANGELOG/README/CLAUDE.md updates, creates a branch if on master, writes a commit message, and commits

    565 GitHub stars~562 tokensUpdated 4 mo ago
    DevelopmentAuto-check passed

More from NVIDIA-NeMo/Curator

  • Repository PR Review Rubric

    NVIDIA-NeMo/Curator

    Supplies the review rubric behind the formal /review command in NeMo Curator: light or strict depth, a narrow focus on real defects and findings filed through the review contract.

    1.8k GitHub stars~447 tokensUpdated today
    Auto-check passed

Works with

Questions about NeMo Curator Docs Maintenance

What does NeMo Curator Docs Maintenance do?

Adds, updates, moves and removes pages on the NeMo Curator Fern documentation site, keeping navigation entries, links and redirects in step. All documentation work happens under `fern/`; the old `docs/` directory is deprecated, and even release notes and migration guides go in the new tree.04`, which `latest` points to through a symlinked config, unless you name another version.

When should I use NeMo Curator Docs Maintenance?

NeMo Curator Docs Maintenance fits situations like: adding a new documentation page to the NeMo Curator Fern site; renaming or moving a page without breaking existing links; removing an outdated page and cleaning up navigation and redirects.

How do I install NeMo Curator Docs Maintenance in Claude Code?

Run `npx skills add NVIDIA-NeMo/Curator --skill nemo-curator-docs -a claude-code`. Or copy the skill folder (.claude/skills/nemo-curator-docs in NVIDIA-NeMo/Curator) into .claude/skills/nemo-curator-docs in your project. Claude Code loads it when a task matches its description.

How do I install NeMo Curator Docs Maintenance in Codex?

Run `npx skills add NVIDIA-NeMo/Curator --skill nemo-curator-docs -a codex`. Or copy the skill folder (.claude/skills/nemo-curator-docs in NVIDIA-NeMo/Curator) into .agents/skills/nemo-curator-docs in your project. Codex loads it when a task matches its description.

Can I use NeMo Curator Docs Maintenance in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add NVIDIA-NeMo/Curator --skill nemo-curator-docs -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/nemo-curator-docs, .gemini/skills/nemo-curator-docs, .github/skills/nemo-curator-docs and .opencode/skills/nemo-curator-docs in your project.

What does NeMo Curator Docs Maintenance need to run?

Going by SKILL.md and its folder, NeMo Curator Docs Maintenance needs the command-line tools its instructions call (git, npm, python and python3) and credentials named FERN_TOKEN. Our summary lists: A checkout of the NeMo Curator repository with its `fern/` docs tree.

Does NeMo Curator Docs Maintenance access the network?

SKILL.md names 2 domains. In commands or code: github.com; the agent is likely to contact it when it follows the instructions. As links in the text: buildwithfern.com. This is read from the text; nothing was executed.

Is NeMo Curator Docs Maintenance safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does NeMo Curator Docs Maintenance use?

NeMo Curator Docs Maintenance is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does NeMo Curator Docs Maintenance use?

About 4.3k tokens (SKILL.md is roughly 17k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to NeMo Curator Docs Maintenance?

Skills that share tags, products or a category with NeMo Curator Docs Maintenance: Release Docs Sync (UfoMiao/zcf, 6.1k stars), Doc Sync (JetBrains/ideavim, 10k stars), Mermaid Diagrams (jjmartres/opencode, 133 stars) and Acquire Codebase Knowledge (github/awesome-copilot, 40k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains NeMo Curator Docs Maintenance?

NVIDIA-NeMo (a GitHub organization) maintains it in NVIDIA-NeMo/Curator, which has 1,802 GitHub stars. The repository holds 2 skills in this directory. The repository was last updated on October 7, 2026.

Source: NVIDIA-NeMo/Curator on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.