feat: add repo-root skills for NVSkills catalog - #104
Merged
Conversation
Publish stable catalog paths for NVSkills CI: evaluator and data-designer plugin skills plus SETUP.md as skills/nemo-setup/SKILL.md. Source-of-truth remains in plugins/ and SETUP.md; this tree is the OSS catalog entrypoint. Signed-off-by: Nick Goncharenko <ngoncharenko@nvidia.com>
Point skills/nemo-evaluator-plugin and skills/nemo-data-designer-plugin at plugin skill trees instead of duplicated copies. Remove root SETUP.md; skills/nemo-setup/SKILL.md is canonical. Update doc links accordingly. Signed-off-by: Nick Goncharenko <ngoncharenko@nvidia.com>
Contributor
|
Contributor
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Enterprise Run ID: 📒 Files selected for processing (1)
✅ Files skipped from review due to trivial changes (1)
📝 WalkthroughWalkthroughThis PR replaces repo-root SETUP.md references with skills/nemo-setup/SKILL.md across docs and skills, and adds comprehensive docs, workflows, references, a persona-schema script, and symlink wiring for the nemo-data-designer-plugin and nemo-evaluator-plugin. ChangesSetup Documentation Consolidation
Data Designer Plugin Documentation
Evaluator Plugin Documentation
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 5
🧹 Nitpick comments (3)
skills/nemo-evaluator-plugin/resources/api-auth.md (1)
1-14: ⚡ Quick winAdd required doc scaffolding (
Prerequisites+Next Steps).This page is missing the required top
Prerequisitessection and endNext Stepscross-links.As per coding guidelines, “Always list prerequisites at the top of documentation pages” and “Include 'Next Steps' section at the end with cross-links to related documentation content.”
🤖 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 `@skills/nemo-evaluator-plugin/resources/api-auth.md` around lines 1 - 14, Add a top-level "Prerequisites" section to skills/nemo-evaluator-plugin/resources/api-auth.md that lists required environment variables, required CLI version, and permissions (e.g., local vs remote distinctions and example like NVIDIA_API_KEY); then append a "Next Steps" section at the end with cross-links to related docs such as the evaluator CLI guide, secrets management (nemo secrets), and remote job submission docs, ensuring each link text and target are explicit so readers can navigate from this page to the evaluator usage and secrets creation guides.skills/nemo-evaluator-plugin/resources/llm-judge.md (1)
1-29: ⚡ Quick winAdd
Prerequisitesat top andNext Stepsat end.Current content is useful, but required doc sections are missing.
As per coding guidelines, “Always list prerequisites at the top of documentation pages” and “Include 'Next Steps' section at the end with cross-links to related documentation content.”
🤖 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 `@skills/nemo-evaluator-plugin/resources/llm-judge.md` around lines 1 - 29, Add a "Prerequisites" section at the top of the "LLM Judge Notes" document (above the existing "LLM Judge Notes" heading) listing required auth/config steps (e.g., Evaluator API auth, model access, CLI installed) and any required permissions or environment variables referenced by the examples (such as workspace/profile). Then add a "Next Steps" section at the end of the file (after the cluster execution example and the advice about --spec-file) that links to related docs like "Evaluator API Auth (api-auth.md)", the CLI reference for `nemo evaluator evaluate` examples, and any onboarding guides; ensure the new headings are titled exactly "Prerequisites" and "Next Steps" so they match site TOC conventions.skills/nemo-evaluator-plugin/resources/troubleshooting.md (1)
1-39: ⚡ Quick winAdd required
PrerequisitesandNext Stepssections.Please add prerequisites before Quick Checks and add end-of-page cross-links.
As per coding guidelines, “Always list prerequisites at the top of documentation pages” and “Include 'Next Steps' section at the end with cross-links to related documentation content.”
🤖 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 `@skills/nemo-evaluator-plugin/resources/troubleshooting.md` around lines 1 - 39, Add a "Prerequisites" section above the "Quick Checks" header that lists required tools and env vars (e.g., nemo CLI installed, access to target workspace/profile, required NeMo platform secrets or local API key env var name) and include a link to the existing "Evaluator API Auth" page (api-auth.md) for auth details; then append a "Next Steps" section at the end of the document that provides cross-links to related docs such as api-auth.md, secrets management (nemo secrets list), and the CLI help pages (nemo evaluator --help and nemo evaluator evaluate --help) so readers can quickly navigate to follow-up actions.
🤖 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.
Inline comments:
In `@skills/nemo-data-designer-plugin/references/person-sampling.md`:
- Around line 40-44: Update the example invocation so it’s unambiguous about
where to run it: either prefix the command with a `cd` into the skill root (so
running `python scripts/get_person_object_schema.py <locale>` from the correct
directory is explicit) or replace the relative path with a repository-root-safe
path to the script; reference the script name get_person_object_schema.py and
the shown command `python scripts/get_person_object_schema.py <locale>` in the
edit so readers understand the exact invocation and required working directory.
In `@skills/nemo-data-designer-plugin/scripts/get_person_object_schema.py`:
- Around line 24-27: Validate the locale string before building
MANAGED_ASSETS_PATH / f"datasets/{locale}.parquet": ensure `locale` matches the
safe pattern (e.g., regex ^[A-Za-z]{2}_[A-Za-z]{2}$) and if not, print a
descriptive error to stderr and exit (same style as the existing error handling)
so path traversal like "../../x" is blocked; perform this check immediately
before the `path = MANAGED_ASSETS_PATH / f"datasets/{locale}.parquet"` line and
reject invalid values referencing the `locale` variable and
`MANAGED_ASSETS_PATH`.
In `@skills/nemo-data-designer-plugin/workflows/autopilot.md`:
- Around line 8-9: The step uses the bare CLI command `nemo data-designer agent
context` which can fail if PATH lookup fails; change the usage to run the
resolved CLI path variable (the same <path> computed in Step 1) so the workflow
always invokes "<path> data-designer agent context" rather than relying on
system PATH; update the instructions and any examples to reference the resolved
path variable consistently and call out that this also applies when falling back
to `.venv/bin/nemo`, keeping existing references like `agent context` and
`model_configs` unchanged.
In `@skills/nemo-data-designer-plugin/workflows/interactive.md`:
- Around line 8-9: Step 2 currently hardcodes the CLI command `nemo
data-designer agent context` and ignores the resolved CLI prefix from Step 1;
update the text to invoke the resolved CLI prefix (the same `<path>`/resolved
binary from Step 1) so the instruction reads `<path> data-designer agent
context` (or equivalent phrasing that instructs users to run the resolved CLI
binary followed by `data-designer agent context`), ensuring it works when only
`.venv/bin/nemo` is available and consistently references the previously
resolved CLI.
In `@skills/nemo-setup/SKILL.md`:
- Around line 87-93: The README line claiming that `make bootstrap` is the
umbrella for three finer-grained targets is incorrect because the table only
lists `make bootstrap-python` and `make bootstrap-studio`; either change the
sentence to say "two finer-grained targets" or add the missing third target
entry to the table (include its target name and one-line “What it does”
description) so the count matches — update the text around `make bootstrap`, and
ensure the table rows reference the exact target names `make bootstrap-python`
and `make bootstrap-studio` (and the new target name if you add one).
---
Nitpick comments:
In `@skills/nemo-evaluator-plugin/resources/api-auth.md`:
- Around line 1-14: Add a top-level "Prerequisites" section to
skills/nemo-evaluator-plugin/resources/api-auth.md that lists required
environment variables, required CLI version, and permissions (e.g., local vs
remote distinctions and example like NVIDIA_API_KEY); then append a "Next Steps"
section at the end with cross-links to related docs such as the evaluator CLI
guide, secrets management (nemo secrets), and remote job submission docs,
ensuring each link text and target are explicit so readers can navigate from
this page to the evaluator usage and secrets creation guides.
In `@skills/nemo-evaluator-plugin/resources/llm-judge.md`:
- Around line 1-29: Add a "Prerequisites" section at the top of the "LLM Judge
Notes" document (above the existing "LLM Judge Notes" heading) listing required
auth/config steps (e.g., Evaluator API auth, model access, CLI installed) and
any required permissions or environment variables referenced by the examples
(such as workspace/profile). Then add a "Next Steps" section at the end of the
file (after the cluster execution example and the advice about --spec-file) that
links to related docs like "Evaluator API Auth (api-auth.md)", the CLI reference
for `nemo evaluator evaluate` examples, and any onboarding guides; ensure the
new headings are titled exactly "Prerequisites" and "Next Steps" so they match
site TOC conventions.
In `@skills/nemo-evaluator-plugin/resources/troubleshooting.md`:
- Around line 1-39: Add a "Prerequisites" section above the "Quick Checks"
header that lists required tools and env vars (e.g., nemo CLI installed, access
to target workspace/profile, required NeMo platform secrets or local API key env
var name) and include a link to the existing "Evaluator API Auth" page
(api-auth.md) for auth details; then append a "Next Steps" section at the end of
the document that provides cross-links to related docs such as api-auth.md,
secrets management (nemo secrets list), and the CLI help pages (nemo evaluator
--help and nemo evaluator evaluate --help) so readers can quickly navigate to
follow-up actions.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: 0f9e31a6-2f4b-4ced-bda4-3126aa6b4b93
📒 Files selected for processing (13)
skills/nemo-data-designer-plugin/SKILL.mdskills/nemo-data-designer-plugin/references/nemo-platform-plugin-additions.mdskills/nemo-data-designer-plugin/references/person-sampling.mdskills/nemo-data-designer-plugin/references/preview-review.mdskills/nemo-data-designer-plugin/references/seed-datasets.mdskills/nemo-data-designer-plugin/scripts/get_person_object_schema.pyskills/nemo-data-designer-plugin/workflows/autopilot.mdskills/nemo-data-designer-plugin/workflows/interactive.mdskills/nemo-evaluator-plugin/SKILL.mdskills/nemo-evaluator-plugin/resources/api-auth.mdskills/nemo-evaluator-plugin/resources/llm-judge.mdskills/nemo-evaluator-plugin/resources/troubleshooting.mdskills/nemo-setup/SKILL.md
This was referenced May 29, 2026
Revert symlinks to plugin trees; catalog source of truth lives at skills/nemo-evaluator-plugin/ and skills/nemo-data-designer-plugin/. Signed-off-by: Nick Goncharenko <ngoncharenko@nvidia.com>
Contributor
There was a problem hiding this comment.
Actionable comments posted: 3
♻️ Duplicate comments (3)
skills/nemo-data-designer-plugin/scripts/get_person_object_schema.py (1)
23-27:⚠️ Potential issue | 🟠 Major | ⚡ Quick winValidate
localebefore constructing the dataset path.Reject invalid locale formats (e.g., traversal patterns) before
MANAGED_ASSETS_PATH / f"datasets/{locale}.parquet".Proposed fix
import sys +import re @@ def main(locale: str) -> None: + if not re.fullmatch(r"[A-Za-z]{2}_[A-Za-z]{2}", locale): + print(f"Error: invalid locale '{locale}'", file=sys.stderr) + sys.exit(1) path = MANAGED_ASSETS_PATH / f"datasets/{locale}.parquet"🤖 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 `@skills/nemo-data-designer-plugin/scripts/get_person_object_schema.py` around lines 23 - 27, Validate the locale string in main() before composing the path with MANAGED_ASSETS_PATH: check that locale is non-empty and contains only an allowed safe pattern (e.g., alphanumeric plus hyphen/underscore, reject path separators and traversal like "/" or ".."), and if it fails print an error to stderr and sys.exit(1) instead of building the path; update the validation logic near main() to run prior to using MANAGED_ASSETS_PATH / f"datasets/{locale}.parquet" and reference the locale variable in the error message for clarity.skills/nemo-data-designer-plugin/workflows/interactive.md (1)
8-37:⚠️ Potential issue | 🟠 Major | ⚡ Quick winUse
<path> data-designerafter CLI resolution, not hardcodednemo ....Step 1 defines a resolved command prefix, but later steps ignore it. Keep all workflow commands aligned with the resolved binary.
Proposed fix
-2. **Learn** — Run `nemo data-designer agent context`. +2. **Learn** — Run `<path> data-designer agent context`. @@ - - If the dataset uses LLM columns, confirm with the user which provider/model(s) to use. Default to declaring `model_configs` programmatically with an IGW provider (portable across local `run` and cluster `submit`); see `references/nemo-platform-plugin-additions.md`. Use `nemo inference providers list` to discover what IGW has registered. + - If the dataset uses LLM columns, confirm with the user which provider/model(s) to use. Default to declaring `model_configs` programmatically with an IGW provider (portable across local `run` and cluster `submit`); see `references/nemo-platform-plugin-additions.md`. Use `<path> inference providers list` to discover what IGW has registered. @@ -6. **Validate** — Run `nemo data-designer validate <path>`. Address any warnings or errors and re-validate until it passes. -7. **Preview** — Run `nemo data-designer preview run <path> --save-results` to generate sample records as HTML files. +6. **Validate** — Run `<path> data-designer validate <path>`. Address any warnings or errors and re-validate until it passes. +7. **Preview** — Run `<path> data-designer preview run <path> --save-results` to generate sample records as HTML files. @@ - - Note the sample records directory printed by the `nemo data-designer preview run` command + - Note the sample records directory printed by the `<path> data-designer preview run` command @@ - - `nemo data-designer create run <path> --num-records <N>`. + - `<path> data-designer create run <path> --num-records <N>`.🤖 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 `@skills/nemo-data-designer-plugin/workflows/interactive.md` around lines 8 - 37, The workflow uses a resolved CLI prefix from Step 1 but later hardcodes "nemo ..."; update all commands to use the resolved prefix variable (e.g., "<path> data-designer" or whatever "resolved command prefix" is named in Step 1) instead of "nemo": replace occurrences like "nemo data-designer validate", "nemo data-designer preview run", "nemo inference providers list", "nemo data-designer create run" and "agent context" invocations with the resolved-prefix form so every command is consistent with the CLI resolution described in Step 1.skills/nemo-data-designer-plugin/workflows/autopilot.md (1)
8-29:⚠️ Potential issue | 🟠 Major | ⚡ Quick winUse the resolved CLI prefix consistently for all commands.
Step 1 says to use
<path> data-designerfor all invocations, but later steps still hardcodenemo data-designer .... This breaks when only.venv/bin/nemois available.Proposed fix
-2. **Learn** — Run `nemo data-designer agent context`. +2. **Learn** — Run `<path> data-designer agent context`. @@ -6. **Validate** — Run `nemo data-designer validate <path>`. Address any warnings or errors and re-validate until it passes. -7. **Preview** — Run `nemo data-designer preview run <path> --save-results` to generate sample records as HTML files. +6. **Validate** — Run `<path> data-designer validate <path>`. Address any warnings or errors and re-validate until it passes. +7. **Preview** — Run `<path> data-designer preview run <path> --save-results` to generate sample records as HTML files. @@ - - Note the sample records directory printed by the `nemo data-designer preview run` command + - Note the sample records directory printed by the `<path> data-designer preview run` command @@ - - Run `nemo data-designer create run <path> --num-records <N>`. + - Run `<path> data-designer create run <path> --num-records <N>`.🤖 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 `@skills/nemo-data-designer-plugin/workflows/autopilot.md` around lines 8 - 29, The docs inconsistently hardcode the CLI as "nemo" in several steps; replace every occurrence of "nemo" with the resolved CLI prefix used in Step 1 (the "<path>" placeholder) so all commands use "<path> data-designer ...". Update the command examples referenced in the file such as "nemo data-designer agent context", "nemo data-designer validate <path>", "nemo data-designer preview run <path> --save-results", and "nemo data-designer create run <path> --num-records <N>" to use the resolved prefix ("<path> data-designer ...") so the guide works when only a venv binary is available.
🧹 Nitpick comments (3)
skills/nemo-data-designer-plugin/references/seed-datasets.md (1)
1-15: ⚡ Quick winAdd a “Next Steps” section with cross-links.
Required by doc standards, and especially useful here to link
person-sampling.md/ workflow docs after seed setup.As per coding guidelines, “Include 'Next Steps' section at the end with cross-links to related documentation content”.
🤖 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 `@skills/nemo-data-designer-plugin/references/seed-datasets.md` around lines 1 - 15, Append a "Next Steps" section to the end of seed-datasets.md that provides cross-links to related docs (e.g., person-sampling.md and the workflow documentation) so readers can continue after configuring a seed; add a brief 2–3 line description and Markdown links to person-sampling.md and the workflow guide (and optionally seed_source.py under the config root) to meet the doc standard requirement.skills/nemo-data-designer-plugin/references/preview-review.md (1)
1-31: ⚡ Quick winAdd required “Prerequisites” (top) and “Next Steps” (end) sections.
This page currently misses two required doc structure elements from repo standards.
As per coding guidelines, “Always list prerequisites at the top of documentation pages” and “Include 'Next Steps' section at the end with cross-links to related documentation content”.
🤖 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 `@skills/nemo-data-designer-plugin/references/preview-review.md` around lines 1 - 31, Add a top-level "Prerequisites" section to the Preview Review Guide that lists required environment, tools, and files (e.g., Python, pandas, path to dataset.parquet or how to find the preview results directory printed as "Results path:"), and add an end "Next Steps" section that links to related docs (examples: Data Designer README, sampler docs, judging/calibration guide) and suggests follow-up actions (run preview, open artifacts, file issues); update the headings in this file (Preview Review Guide) so the new "Prerequisites" appears above "Mindset" and the new "Next Steps" appears after "Design Choices" with clear cross-links and brief actionable items.skills/nemo-evaluator-plugin/SKILL.md (1)
22-22: ⚡ Quick winReplace hardcoded venv activation with
uv-based invocation.
source .venv/bin/activateassumes env layout and can fail in standard repo workflows. Prefer commands that work directly withuv(for example,uv run nemo ...) in this skill guidance.Based on learnings: "Use uv exclusively for Python package management in all projects; never use pip, pip-tools, poetry, or conda directly" and "Run Python scripts with
uv run <script-name>.py, run tools like Pytest withuv run pytest, and launch Python repl withuv run python."🤖 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 `@skills/nemo-evaluator-plugin/SKILL.md` at line 22, Replace the hardcoded virtualenv activation line that reads "source .venv/bin/activate" in SKILL.md with a uv-based invocation; update the "Prerequisite" guidance to instruct users to run the skill via the uv tool (e.g., "uv run nemo ..." or "uv run python -m nemo" as appropriate) so the documentation uses uv for running the nemo CLI and other Python commands instead of assuming .venv layout.
🤖 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.
Inline comments:
In `@skills/nemo-data-designer-plugin/workflows/autopilot.md`:
- Around line 5-29: The ordered list in the Autopilot workflow uses sequential
numbering instead of the required MD029-consistent style; update every top-level
numbered item (e.g., the lines starting "Resolve CLI command", "Learn", "Infer",
"Plan", "Build", "Validate", "Preview", "Create", "Present") to use the same
ordered-list prefix (preferably "1.") so markdownlint MD029 is satisfied and
list rendering stays consistent.
In `@skills/nemo-data-designer-plugin/workflows/interactive.md`:
- Around line 5-33: The ordered list in the workflow (the steps beginning with
"1. Resolve CLI command — Run `command -v nemo ...`" through "9. **Finalize** —
Once the user is happy...") mixes numeric prefixes and triggers markdownlint
MD029; normalize the list prefixes to the repository's configured ordered-list
style (e.g., make every top-level list item use the same marker such as "1." or
use strictly sequential numbers) so all nine steps use a consistent numeric
prefix style and nested bullets remain unchanged.
In `@skills/nemo-evaluator-plugin/resources/api-auth.md`:
- Around line 11-13: Replace the CLI example that passes the secret via argv
(nemo secrets create nvidia-api-key --value "$NVIDIA_API_KEY") with a non-argv
approach: either show creating the secret via the Platform SDK call
sdk.secrets.create(name="nvidia-api-key", data=...) (include how to pass the
variable as data) or demonstrate a CLI flow that reads the secret from stdin or
a file (e.g., piping the env var into nemo secrets create --stdin or using a
file and a --from-file style flag) so the secret is not exposed in process
argv/history; update the example in api-auth.md to use sdk.secrets.create(...,
data=...) or a documented stdin/file CLI variant and remove the --value usage.
---
Duplicate comments:
In `@skills/nemo-data-designer-plugin/scripts/get_person_object_schema.py`:
- Around line 23-27: Validate the locale string in main() before composing the
path with MANAGED_ASSETS_PATH: check that locale is non-empty and contains only
an allowed safe pattern (e.g., alphanumeric plus hyphen/underscore, reject path
separators and traversal like "/" or ".."), and if it fails print an error to
stderr and sys.exit(1) instead of building the path; update the validation logic
near main() to run prior to using MANAGED_ASSETS_PATH /
f"datasets/{locale}.parquet" and reference the locale variable in the error
message for clarity.
In `@skills/nemo-data-designer-plugin/workflows/autopilot.md`:
- Around line 8-29: The docs inconsistently hardcode the CLI as "nemo" in
several steps; replace every occurrence of "nemo" with the resolved CLI prefix
used in Step 1 (the "<path>" placeholder) so all commands use "<path>
data-designer ...". Update the command examples referenced in the file such as
"nemo data-designer agent context", "nemo data-designer validate <path>", "nemo
data-designer preview run <path> --save-results", and "nemo data-designer create
run <path> --num-records <N>" to use the resolved prefix ("<path> data-designer
...") so the guide works when only a venv binary is available.
In `@skills/nemo-data-designer-plugin/workflows/interactive.md`:
- Around line 8-37: The workflow uses a resolved CLI prefix from Step 1 but
later hardcodes "nemo ..."; update all commands to use the resolved prefix
variable (e.g., "<path> data-designer" or whatever "resolved command prefix" is
named in Step 1) instead of "nemo": replace occurrences like "nemo data-designer
validate", "nemo data-designer preview run", "nemo inference providers list",
"nemo data-designer create run" and "agent context" invocations with the
resolved-prefix form so every command is consistent with the CLI resolution
described in Step 1.
---
Nitpick comments:
In `@skills/nemo-data-designer-plugin/references/preview-review.md`:
- Around line 1-31: Add a top-level "Prerequisites" section to the Preview
Review Guide that lists required environment, tools, and files (e.g., Python,
pandas, path to dataset.parquet or how to find the preview results directory
printed as "Results path:"), and add an end "Next Steps" section that links to
related docs (examples: Data Designer README, sampler docs, judging/calibration
guide) and suggests follow-up actions (run preview, open artifacts, file
issues); update the headings in this file (Preview Review Guide) so the new
"Prerequisites" appears above "Mindset" and the new "Next Steps" appears after
"Design Choices" with clear cross-links and brief actionable items.
In `@skills/nemo-data-designer-plugin/references/seed-datasets.md`:
- Around line 1-15: Append a "Next Steps" section to the end of seed-datasets.md
that provides cross-links to related docs (e.g., person-sampling.md and the
workflow documentation) so readers can continue after configuring a seed; add a
brief 2–3 line description and Markdown links to person-sampling.md and the
workflow guide (and optionally seed_source.py under the config root) to meet the
doc standard requirement.
In `@skills/nemo-evaluator-plugin/SKILL.md`:
- Line 22: Replace the hardcoded virtualenv activation line that reads "source
.venv/bin/activate" in SKILL.md with a uv-based invocation; update the
"Prerequisite" guidance to instruct users to run the skill via the uv tool
(e.g., "uv run nemo ..." or "uv run python -m nemo" as appropriate) so the
documentation uses uv for running the nemo CLI and other Python commands instead
of assuming .venv layout.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: 72f1daba-35de-40da-9381-a1a74067f399
📒 Files selected for processing (12)
skills/nemo-data-designer-plugin/SKILL.mdskills/nemo-data-designer-plugin/references/nemo-platform-plugin-additions.mdskills/nemo-data-designer-plugin/references/person-sampling.mdskills/nemo-data-designer-plugin/references/preview-review.mdskills/nemo-data-designer-plugin/references/seed-datasets.mdskills/nemo-data-designer-plugin/scripts/get_person_object_schema.pyskills/nemo-data-designer-plugin/workflows/autopilot.mdskills/nemo-data-designer-plugin/workflows/interactive.mdskills/nemo-evaluator-plugin/SKILL.mdskills/nemo-evaluator-plugin/resources/api-auth.mdskills/nemo-evaluator-plugin/resources/llm-judge.mdskills/nemo-evaluator-plugin/resources/troubleshooting.md
✅ Files skipped from review due to trivial changes (5)
- skills/nemo-evaluator-plugin/resources/llm-judge.md
- skills/nemo-data-designer-plugin/references/person-sampling.md
- skills/nemo-evaluator-plugin/resources/troubleshooting.md
- skills/nemo-data-designer-plugin/references/nemo-platform-plugin-additions.md
- skills/nemo-data-designer-plugin/SKILL.md
Signed-off-by: Nick Goncharenko <ngoncharenko@nvidia.com>
3 tasks
gabwow
approved these changes
May 29, 2026
Signed-off-by: Nick Goncharenko <ngoncharenko@nvidia.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Publish NeMo Platform skills under stable
skills/paths for the NVSkills catalog. Evaluator, data-designer, and setup content live inskills/(not symlinks into plugin trees). RootSETUP.mdis removed in favor ofskills/nemo-setup/SKILL.md.Why
The NVSkills contributing guide recommends
skills/at repo root for OSS catalog entrypoints. Shallow paths keep harvesting predictable; keeping real files underskills/makes the catalog path the source of truth.nemo-evaluator-pluginskills/nemo-evaluator-plugin/nemo-data-designer-pluginskills/nemo-data-designer-plugin/nemo-setupskills/nemo-setup/SKILL.mdChanges
skills/nemo-evaluator-plugin/— evaluator plugin skill + resourcesskills/nemo-data-designer-plugin/— data designer skill + references, workflows, scriptsskills/nemo-setup/SKILL.md— canonical setup guide (replaces rootSETUP.md)Related
NVIDIA/skillscomponents.d/nemo-platform.yml→skills/**pathsTest plan
tree skillsshows real directories (no symlinks toplugins/)SKILL.mdand relative links resolve/nvskills-cion a PR including this branchSummary by CodeRabbit