docs: fix stale links and skill listing

[glenn-agent]

Summary

Fix several low-risk documentation navigation and discoverability issues reported by doc validation:

• Fix stale links and version placeholders in the commands reference.
• Point architecture and release-note links at the existing Sandbox Hardening, Windows preparation, and Hermes quickstart pages.
• Add the missing running-sandbox prerequisite to the sub-agent setup guide.
• List nemoclaw-user-agent-skills in the agent skills page and add an example question for it.
• Fix broken Windows and Hermes links in troubleshooting.

Fixes #5445
Fixes #5089
Fixes #5079
Fixes #5077
Fixes #5083
Fixes #5076

Testing

• npm run docs:sync-agent-variants
• npm run docs:strict — passed with 0 errors and 2 Fern upgrade warnings
• git diff --check
• npm run docs:check-agent-variants
• git verify-commit HEAD

Signed-off-by: Glenn-Agent glenn_agent@163.com

Summary by CodeRabbit

Documentation

• Fixed internal documentation links across setup and reference guides for improved navigation accuracy
• Updated version placeholders in command examples to use generic format for broader compatibility
• Added prerequisite requirements and context to sub-agent setup instructions for clarity
• Expanded skill documentation to include the NemoClaw user agent skills feature with example usage

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[coderabbitai]

📝 Walkthrough

Walkthrough

Fixes broken internal documentation links across six pages (how-it-works, release-notes, troubleshooting, commands reference), replaces hardcoded agent version strings with generic vX.Y.Z placeholders in command reference examples, adds prerequisite guidance to the sub-agent setup page, and adds the missing nemoclaw-user-agent-skills entry to the agent skills reference.

Changes

Documentation Link Fixes, Version Placeholders, and Content Additions

Layer / File(s)	Summary
Internal link corrections docs/about/how-it-works.mdx, docs/about/release-notes.mdx, docs/reference/troubleshooting.mdx, docs/reference/commands.mdx	Corrects broken or stale relative link targets: sandbox-hardening and Hermes quickstart in how-it-works, Windows bootstrap and Hermes quickstart in release-notes, Windows preparation (×2) and Hermes quickstart (×2) in troubleshooting, and OpenClaw plugin install page in commands.
Version string generalization docs/reference/commands.mdx, docs/reference/commands-nemohermes.mdx	Replaces pinned version strings (v2026.5.27, v2026.5.16) in expected-output code blocks with generic vX.Y.Z placeholders for both OpenClaw and Hermes.
Prerequisite note and new skill entry docs/inference/set-up-sub-agent.mdx, docs/resources/agent-skills.mdx	Inserts a prerequisite section before the Docker exec commands in set-up-sub-agent.mdx, and adds nemoclaw-user-agent-skills to the Available Skills table and example questions mapping in agent-skills.mdx.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~5 minutes

Possibly related PRs

• NVIDIA/NemoClaw#5247: Modifies the same "Update the Sandbox Config" section in docs/inference/set-up-sub-agent.mdx — directly overlaps with this PR's prerequisite addition in that section.
• NVIDIA/NemoClaw#5358: Updates link targets in docs/about/release-notes.mdx for Hermes and Windows paths, the same links corrected here.

Suggested labels

area: docs, area: skills

Suggested reviewers

• cv
• prekshivyas

Poem

🐇 Hoppity-hop through the docs I go,
Fixing the links that led to "404 no!"
Version strings swapped to vX.Y.Z,
A skill was lost — now found for thee!
The prereqs are clear, the paths are right,
This bunny fixed docs in the moonlight. ✨

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'docs: fix stale links and skill listing' accurately summarizes the main changes—fixing broken documentation links and adding missing skill entries—making it clear and directly related to the changeset.
Linked Issues check	✅ Passed	All six linked issues are fully addressed: hardcoded versions and broken plugin link fixed (#5445), sandbox-hardening and hermes quickstart paths corrected (#5089), Windows and Hermes links updated in release notes (#5079), prerequisite added to sub-agents setup (#5077), nemoclaw-user-agent-skills added to skills page (#5083), and troubleshooting links fixed (#5076).
Out of Scope Changes check	✅ Passed	All changes are documentation-only and directly address the six linked issues; no unrelated modifications or scope creep detected in the changeset.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

🧹 Nitpick comments (1)

docs/inference/set-up-sub-agent.mdx (1)

55-55: ⚡ Quick win

Use active voice here.

Line 55 reads passively. As per coding guidelines, use active voice and second person in docs.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@docs/inference/set-up-sub-agent.mdx` at line 55, Rewrite line 55 in
docs/inference/set-up-sub-agent.mdx to use active voice and second person
perspective instead of passive voice. The current sentence uses passive
constructions like "must be running" and "must be discoverable". Convert it to
direct, imperative instructions that address the reader directly using "you" and
"your", such as by making it clear what actions the reader must take to ensure
the sandbox is properly configured and discoverable.

Source: Coding guidelines

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Nitpick comments:
In `@docs/inference/set-up-sub-agent.mdx`:
- Line 55: Rewrite line 55 in docs/inference/set-up-sub-agent.mdx to use active
voice and second person perspective instead of passive voice. The current
sentence uses passive constructions like "must be running" and "must be
discoverable". Convert it to direct, imperative instructions that address the
reader directly using "you" and "your", such as by making it clear what actions
the reader must take to ensure the sandbox is properly configured and
discoverable.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: e0f9a93c-ca7b-4d58-bdaa-f600a823c596

📥 Commits

Reviewing files that changed from the base of the PR and between f4f3c58 and d809dbc.

📒 Files selected for processing (7)

• docs/about/how-it-works.mdx
• docs/about/release-notes.mdx
• docs/inference/set-up-sub-agent.mdx
• docs/reference/commands-nemohermes.mdx
• docs/reference/commands.mdx
• docs/reference/troubleshooting.mdx
• docs/resources/agent-skills.mdx

[miyoungc]

Thanks for taking the time to work through these documentation findings.

We are closing this PR because the part that applies to #5083 has already been resolved and merged in #5522. main now includes the nemoclaw-user-agent-skills row in the Available Skills table and the matching example question, so there is nothing left from #5083 that this PR needs to carry.

The remaining link changes in this PR should not be merged as-is. Several of those QA findings came from resolving Fern docs links as raw filesystem paths, but NemoClaw's published docs use route-style links from docs/index.yml. Converting those links to source-file paths would risk breaking the published Fern routes.

Appreciate the contribution. If there is a separate unresolved docs issue, please open a fresh focused PR against that specific issue.