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

Inline comments:
In
@.agents/skills/nemoclaw-user-configure-security/references/best-practices.md:
- Line 5: The change was made directly in the generated skill artifact
(.agents/skills/nemoclaw-user-configure-security/references/best-practices.md);
revert any manual edits to that generated file, update the canonical source
documentation instead, then run the documented docs-to-skills regeneration step
to rebuild the nemoclaw-user-configure-security skill so the deny-by-default
security wording is produced from source and the generated artifact remains
reproducible and consistent.

In @.agents/skills/nemoclaw-user-overview/references/release-notes.md:
- Line 47: Update the autogenerated release notes entry that currently has
multiple sentence-like clauses on one line: break the long bullet "Release
validation now runs real shell-boundary assertions through Vitest E2E support,
includes an opt-in live scenario project, shards CLI coverage, adds a docs-only
PR fast path, and trims slow CLI subprocess coverage." into separate lines so
each sentence/clause is on its own line in release-notes.md, and then modify the
generation script that produces this content to emit one sentence per line
(preserve punctuation and ordering).
- Around line 13-16: Split each bullet into one sentence per line by breaking
the multi-sentence lines into separate lines: for the bullet starting "Sandbox
lifecycle commands preserve and recover more state." put each sentence on its
own line, do the same for the bullets beginning "Snapshot-backed rebuilds
preserve OpenClaw configuration more safely.", "Onboarding diagnoses host setup
and local inference issues earlier.", and "Messaging and Hermes startup paths
enforce clearer runtime boundaries."; update the generator that emits
release-notes.md so it joins individual sentences with newline separators
(ensuring punctuation-aware sentence splitting) instead of emitting multiple
sentences on a single line, and regenerate the file so each sentence appears on
its own line.

In @.agents/skills/nemoclaw-user-reference/references/troubleshooting.md:
- Around line 91-104: This file
(.agents/skills/nemoclaw-user-reference/references/troubleshooting.md) contains
edits to autogenerated skill content that must not be modified manually; revert
these manual changes and regenerate the skill from the canonical docs generation
workflow so the content is rebuilt from source (i.e., run the
documentation/skill generation script or pipeline that produces the
nemoclaw-user-* skills), ensuring all related places (the other autogenerated
ranges mentioned) are updated consistently rather than editing
troubleshooting.md by hand.

In `@docs/about/release-notes.mdx`:
- Around line 22-25: Split each bullet into one sentence per line for
readability: break the first bullet that starts "Sandbox lifecycle commands
preserve and recover more state." into three separate lines (one for the
`rebuild --yes` sentence, one for the Docker-driver sandboxes sentence, and one
for the `upgrade-sandboxes` sentence), break the second bullet that starts
"Snapshot-backed rebuilds preserve OpenClaw configuration more safely." into
three lines (one for carrying forward `openclaw.json`, one for merging restored
config, and one for failing when config cannot be applied), break the third
bullet that starts "Onboarding diagnoses host setup and local inference issues
earlier." into four lines (one per clause about Docker group, host DNS, Ollama
auth-proxy port conflicts, and managed vLLM model picker), and break the fourth
bullet that starts "Messaging and Hermes startup paths enforce clearer runtime
boundaries." into two lines (one for Slack setup validation and one for Hermes
direct gateway launch) so each sentence sits on its own line.

---

Outside diff comments:
In
@.agents/skills/nemoclaw-user-configure-security/references/best-practices.md:
- Line 1: Add the required SPDX copyright and license header as HTML comments at
the very top of the Markdown file (before the title line "# NemoClaw Security
Best Practices: Controls, Risks, and Posture Profiles"); include the SPDX short
identifier (e.g., "SPDX-FileCopyrightText" and "SPDX-License-Identifier") per
the coding guidelines so the file contains the standard SPDX header block in
HTML comment form.

In @.agents/skills/nemoclaw-user-overview/references/release-notes.md:
- Line 1: Add the required SPDX HTML comment header to autogenerated Markdown by
updating the generator in scripts/docs-to-skills.py so it prepends the two lines
<!-- SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES.
All rights reserved. --> and <!-- SPDX-License-Identifier: Apache-2.0 --> to
output files (e.g., the generated release-notes.md in
.agents/skills/nemoclaw-user-overview/references/); modify the markdown-emitting
function (the routine that writes or returns the release notes content) to
insert these lines at the top before writing the file so all generated .md files
include the SPDX header.

In @.agents/skills/nemoclaw-user-reference/references/commands.md:
- Line 1: Add the required SPDX header HTML comments to the top of the Markdown
file by inserting an SPDX copyright comment and an SPDX license identifier
comment as HTML comments before the existing content in
.agents/skills/nemoclaw-user-reference/references/commands.md (e.g., add an <!--
SPDX-FileCopyrightText: ... --> line and an <!-- SPDX-License-Identifier: ...
--> line immediately above the "# NemoClaw CLI Commands Reference" heading).

In @.agents/skills/nemoclaw-user-reference/references/troubleshooting.md:
- Line 1: Prepend the required SPDX HTML comment header to the top of the
Markdown file by adding a single HTML comment that includes the
SPDX-License-Identifier (e.g., "SPDX-License-Identifier: MIT") and the
copyright/ownership line as required by our guidelines; ensure this comment is
the very first content in troubleshooting.md so the file complies with the "All
source files must include an SPDX license header" rule.

---

Nitpick comments:
In `@docs/manage-sandboxes/backup-restore.mdx`:
- Line 84: Rewrite the sentence in active voice so NemoClaw is the subject;
e.g., change "If the restored config cannot be parsed or applied safely,
NemoClaw stops the restore..." to "If NemoClaw cannot safely parse or apply the
restored config, it stops the restore instead of replacing the generated config
with an unsafe fallback." Update the sentence in the
docs/manage-sandboxes/backup-restore.mdx where the current line about the
restored config and NemoClaw's behavior appears.
- Line 83: The sentence beginning "During rebuild or restore, NemoClaw merges
those restored settings with the freshly generated runtime config..." is long
and uses the informal phrase "win over"; split it into two clearer sentences and
replace "win over" with a precise term like "take precedence over" or "override"
so it reads e.g. "During rebuild or restore, NemoClaw merges restored settings
with the freshly generated runtime config. Current provider placeholders,
messaging enablement, and gateway state take precedence over stale snapshot
values." Ensure punctuation and flow match surrounding documentation.

In `@docs/manage-sandboxes/lifecycle.mdx`:
- Line 240: Rewrite the passive sentence into active voice: change "Custom-image
sandboxes created with `--from <Dockerfile>` are not marked stale solely by
image fingerprint, so an upgrade check does not accidentally replace them with
the default image." to an active construction that clearly states who/what
performs the action (e.g., "The system does not mark custom-image sandboxes
created with `--from <Dockerfile>` as stale solely by image fingerprint, so an
upgrade check does not accidentally replace them with the default image.").
Ensure the meaning and flags (`--from <Dockerfile>`) remain unchanged.

In `@docs/reference/commands-nemohermes.mdx`:
- Line 1091: The sentence "Custom-image sandboxes created with `--from
<Dockerfile>` are not classified by image drift because rebuilding them onto the
default image would drop the custom image." uses passive voice; rewrite it in
active voice in docs/reference/commands-nemohermes.mdx by making the subject
perform the action (e.g., "Nemo Hermes does not classify custom-image sandboxes
created with `--from <Dockerfile>` for image drift because rebuilding them onto
the default image would drop the custom image.") — update the sentence to this
active-voice form or a similar active construction while keeping the meaning
unchanged.

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

Inline comments:
In @.agents/skills/nemoclaw-contributor-update-docs/SKILL.md:
- Line 205: The sentence "For post-release documentation refreshes, label the PR
with the next patch release label" is ambiguous relative to the other references
to "next-patch release label" later in the doc; update the phrasing so the
policy is consistent: either (A) state explicitly that the "next-patch" label
rule applies to all release-prep work (replace the conditional "For
post-release..." in the sentence with an unconditional rule) or (B) state that
the "next-patch" label applies only to post-release refreshes and update the
other occurrences that refer to "next-patch release label" (the paragraphs that
currently read "next-patch release label") to instead instruct using the
documented version label for pre-release work; ensure you keep the example about
0.0.63→v0.0.64 and add a brief note to ask when the version is not provided or
is nonstandard.

---

Outside diff comments:
In @.agents/skills/nemoclaw-contributor-update-docs/SKILL.md:
- Line 216: The heading "Step 9: Build and Verify" should be renumbered to "Step
8: Build and Verify" to fix the skipped step; update the markdown heading text
and any nearby references or anchors that mention "Step 9" (search for the
literal "Step 9: Build and Verify" and change to "Step 8: Build and Verify") so
it follows "Step 7: Apply Release Prep Updates" consistently and ensure any
cross-references or table of contents entries are adjusted accordingly.