Skip to content
Back to search

nebius-physical-ai

The control plane your coding agent uses to run physical-AI workloads on Nebius.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/nebius/nebius-physical-ai with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/nebius/nebius-physical-ai'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/nebius--nebius-physical-ai/badge.svg)](/listings/nebius--nebius-physical-ai)

HTML

<a href="/listings/nebius--nebius-physical-ai"><img src="/api/listings/nebius--nebius-physical-ai/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Using a coding agent? Open this checkout in your agent and paste one of the maintained prompts:

- [Set up Workbench for a task](docs/workbench/agent-first-run.md#set-up-workbench). - [Run the Cosmos 3 generation workflow](docs/workbench/agent-first-run.md#run-cosmos-3-generation).

The prompts keep credentials private, stop at human approval gates, validate the plan before provisioning, and stay with the run through artifact inspection. The manual path below exposes the same control-plane steps.

1. Install

Use Python **3.10+** on macOS, Linux, or WSL2 Ubuntu. Clone the repository, then run the remaining commands from its root:

git clone https://github.com/nebius/nebius-physical-ai.git
cd nebius-physical-ai
python3 -m venv .venv
source .venv/bin/activate
pip install -e npa
npa --version

You should see the installed `npa` version. Remote GPU workloads use their container dependencies; you do not need CUDA or a local GPU to use the CLI. [Installation](docs/install.md) covers platform setup and optional dependencies.

2. Connect your project

Install the tested Nebius CLI and configure your project:

curl -fsSL https://storage.eu-north1.nebius.cloud/cli/install.sh \
  | NEBIUS_CLI_VERSION=0.12.254 bash
export PATH="${HOME}/.nebius/bin:${PATH}"
npa configure
npa workbench health preflight --checks nebius --json

The preflight should report working Nebius authentication. It does not reserve GPUs or verify model access. Interactive configuration provisions object storage by default and needs project admin permission for that setup. Use `npa configure --no-provision` to save settings without creating storage. See [configuration](docs/configuration.md) for SSO, existing projects, and tokens.

Check a project's saved storage with its configured alias in place of `<alias>`:

npa workbench health preflight --project '<alias>' --checks s3,nebius --json

Missing or invalid project storage fails without borrowing shell or host-file credentials or changing configuration. S3 readiness uses a bounded listing call; it does not enumerate the dataset or verify write access. Without `--project`, the S3 check keeps the existing host credential selection.

3. Run and inspect

Choose one guide below and follow it through input preparation, GPU setup, submission, and output inspection. [Workbench setup](docs/workbench/getting-started.md) provides the shared cluster and SkyPilot steps; complete them once per project. Keep the returned run ID: you use it to inspect logs, find artifacts, and resume.

For video augmentation, the [PAIDF + Cosmos 3 runbook](workflows/guides/paidf-cosmos3.md) includes a public source-video example and full submission commands.

<a id="pick-your-first-win"></a>

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 359 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. <div align="center">
  2. # Nebius Physical AI
  3. **The agent control plane for Physical AI on Nebius.**
  4. <img src="docs/assets/workbench-architecture.png" alt="Workbench architecture: selected NVIDIA and open-ecosystem integrations are packaged into public GHCR images. npa uses workflow YAML, its workflow engine, and SkyPilot to run workloads on Nebius Kubernetes. Kubernetes pulls images; Object Storage holds artifacts and run state. Token Factory provides hosted inference. Models and vendor runtimes are fetched at runtime where required." width="960" />
  5. [![License: Apache 2.0](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](LICENSE)
  6. [![Python 3.11+](https://img.shields.io/badge/python-3.11%2B-blue.svg)](https://www.python.org/downloads/)
  7. [![Platforms: macOS · Linux · WSL2](https://img.shields.io/badge/platforms-macOS%20%7C%20Linux%20%7C%20WSL2-lightgrey.svg)](docs/install.md)
  8. [![Test](https://github.com/nebius/nebius-physical-ai/actions/workflows/test.yml/badge.svg)](https://github.com/nebius/nebius-physical-ai/actions/workflows/test.yml)
  9. [![PRs welcome](https://img.shields.io/badge/PRs-welcome-brightgreen.svg)](CONTRIBUTING.md)
  10. **[Quickstart](docs/quickstart.md)** ·
  11. **[Guides](docs/workbench/guides/README.md)** ·
  12. **[Manual workflows](docs/workbench/guides/manual-workflow-operations.md)** ·
  13. **[Workbench docs](docs/workbench/README.md)** ·
  14. **[Shared access](docs/workbench/shared-workbench-quickstart.md)** ·
  15. **[Benchmarks](benchmark/README.md)** ·
  16. **[Operator tools](docs/tools/README.md)** ·
  17. **[CLI reference](docs/cli/README.md)** ·
  18. **[Python & API](docs/workbench/cli-sdk-yaml-walkthrough.md)** ·
  19. **[Cookbooks](docs/workbench/cookbooks/README.md)** ·
  20. **[Contributing](CONTRIBUTING.md)**
  21. </div>
  22. ## Demos
  23. **Video augmentation · 8 seconds**
  24. One ALOHA recording, twelve Cosmos3 appearance variants. This batch did not pass
  25. its quality gate. [Workflow](workflows/main/paidf-cosmos3.yaml) ·
  26. [Runbook](workflows/guides/paidf-cosmos3.md) ·
  27. [MP4](https://github.com/user-attachments/assets/95917c76-7920-420f-af08-c13590a36a3f)
  28. <img src="https://github.com/user-attachments/assets/5df4f2c0-c213-4662-a892-5856e90c8269" alt="One ALOHA recording alongside twelve Cosmos3 appearance variants, looping automatically." width="640" />
  29. **Robot locomotion · 8 seconds**
  30. A Unitree G1 policy rollout in MJLab simulation.
  31. [Workflow](workflows/testing/mjlab-render.yaml) ·
  32. [MP4](https://github.com/user-attachments/assets/7e605a6d-b32b-4746-808c-51258a11b21a)
  33. <img src="https://github.com/user-attachments/assets/34e720b9-df15-43a6-b14d-680e01ce4ee4" alt="A Unitree G1 walking in MJLab simulation, looping automatically." width="640" />
  34. **Video segmentation · 6 seconds**
  35. SAM 2.1 follows an object through generated video.
  36. [Workflow](workflows/testing/byof-sam2.1.yaml) ·
  37. [MP4](https://github.com/user-attachments/assets/f629401a-346d-4862-b221-34b5ecc28128)
  38. <img src="https://github.com/user-attachments/assets/e4edfa25-ce1a-4ab9-a745-8d3680d86967" alt="SAM 2.1 tracking a generated mobile robot with a colored mask, looping automatically." width="640" />
  39. [Clip sources and scope](docs/demos/readme-videos.md)
  40. For a shared Workbench endpoint, start with the
  41. [shared-access quickstart](docs/workbench/shared-workbench-quickstart.md).
  42. Operators install the HTTPS service once; users and coding agents connect with
  43. `npa login` using a personal key or Nebius human sign-in. Workbench access adds
  44. personal namespaces, GPU caps, and scoped storage. The CLI and agents reuse
  45. the saved connection through the same authenticated API.
  46. Operators managing Kubernetes directly can use
  47. [team namespaces](docs/workbench/namespaces.md) to select namespaces and private
  48. SkyPilot contexts with `npa workbench namespace`.
  49. ## What is Workbench?
  50. Workbench is NPA's agent control plane for Physical AI on Nebius AI Cloud,
  51. covering infrastructure setup and management, simulation, robot policy training,
  52. synthetic data, and evaluation. Codex, Claude Code, or another coding agent with
  53. terminal access to this checkout uses repository skills and `npa` to configure
  54. projects, check access, provision GPU clusters and object storage, run
  55. containerized tools and workflows, and inspect artifacts with you. You can use
  56. the same operations directly through the CLI; supported tools also expose
  57. Python interfaces.
  58. ### How Workbench runs a task
  59. The architecture overview maps selected NVIDIA and open-ecosystem solutions to
  60. NPA's [public GHCR container catalog](docs/workbench/container-image-catalog.md)
  61. at `ghcr.io/nebius/nebius-physical-ai` and the Nebius services that run and support
  62. them. Some images fetch models or vendor runtimes at execution time; the
  63. [packaging contract](docs/workbench/container-packaging.md) records those boundaries.
  64. For the standard Kubernetes workflow path, the
  65. [workflow engine](docs/workbench/npa-workflow-guide.md#runtime-orchestrator---runtime)
  66. plans execution waves, submits them through SkyPilot, and uses S3 artifacts and
  67. durable run state to evaluate decisions and resume runs. The operator-side
  68. [SkyPilot environment](docs/orchestration/skypilot-setup.md) submits the jobs;
  69. the workload containers run on Nebius GPU or CPU nodes.
  70. [Token Factory](docs/workbench/token-factory.md) is a separate hosted inference
  71. API. Direct calls need no cluster; workflow stages can also call it from their
  72. containers. Individual tools additionally support the
  73. [runtime modes documented for that tool](docs/workbench/README.md).
  74. Inspect the resulting artifacts with Rerun, Foxglove, media viewers, or reports.
  75. You remain in the loop for choices and human-bound approvals such as accepting
  76. model terms. The [diagram reference](docs/assets/workbench-architecture.md)
  77. records the scope and source files behind the visuals.
  78. Start with a [workload guide](#pick-your-first-win). The guide tells you which
  79. data, model access, GPU, and output to expect.
  80. ## Quickstart
  81. Using a coding agent? Open this checkout in your agent and paste one of the
  82. maintained prompts:
  83. - [Set up Workbench for a task](docs/workbench/agent-first-run.md#set-up-workbench).
  84. - [Run the Cosmos 3 generation workflow](docs/workbench/agent-first-run.md#run-cosmos-3-generation).
  85. The prompts keep credentials private, stop at human approval gates, validate
  86. the plan before provisioning, and stay with the run through artifact inspection.
  87. The manual path below exposes the same control-plane steps.
  88. Joining an existing shared Workbench? Complete **1. Install**, then follow the
  89. [shared-access quickstart](docs/workbench/shared-workbench-quickstart.md).
  90. The project connection steps below are for operators; `npa configure` does not
  91. install the shared service or its LoadBalancer.
  92. ### 1. Install
  93. Use Python **3.11+** on macOS, Linux, or WSL2 Ubuntu. Clone the repository,
  94. then run the remaining commands from its root:
  95. ```bash
  96. git clone https://github.com/nebius/nebius-physical-ai.git
  97. cd nebius-physical-ai
  98. python3 -m venv .venv
  99. source .venv/bin/activate
  100. pip install -e npa
  101. npa --version
  102. ```
  103. You should see the installed `npa` version. Remote GPU workloads use their
  104. container dependencies; you do not need CUDA or a local GPU to use the CLI.
  105. [Installation](docs/install.md) covers platform setup and optional dependencies.
  106. ### 2. Connect your project
  107. Install the tested Nebius CLI and configure your project:
  108. ```bash
  109. curl -fsSL https://storage.eu-north1.nebius.cloud/cli/install.sh \
  110. | NEBIUS_CLI_VERSION=0.12.298 bash
  111. export PATH="${HOME}/.nebius/bin:${PATH}"
  112. npa configure
  113. npa workbench health preflight --checks nebius --json
  114. ```
  115. The preflight should report working Nebius authentication. It does not reserve
  116. GPUs or verify model access. Interactive configuration provisions object storage
  117. by default and needs project admin permission for that setup. Use
  118. `npa configure --no-provision` to save settings without creating storage.
  119. See [configuration](docs/configuration.md) for SSO, existing projects, and tokens.
  120. Check a project's saved storage with its configured alias in place of `<alias>`:
  121. ```bash
  122. npa workbench health preflight --project '<alias>' --checks s3,nebius --json
  123. ```
  124. Missing or invalid project storage fails without borrowing shell or host-file
  125. credentials or changing configuration. S3 readiness uses a bounded listing call;
  126. it does not enumerate the dataset or verify write access. Without `--project`,
  127. the S3 check keeps the existing host credential selection.
  128. ### 3. Run and inspect
  129. Choose one guide below and follow it through input preparation, GPU setup,
  130. submission, and output inspection. [Workbench setup](docs/workbench/getting-started.md)
  131. provides the shared cluster and SkyPilot steps; complete them once per project.
  132. Keep the returned run ID: you use it to inspect logs, find artifacts, and resume.
  133. For video augmentation, the [PAIDF + Cosmos 3 runbook](workflows/guides/paidf-cosmos3.md)
  134. includes a public source-video example and full submission commands.
  135. For a shared installation, [team PAIDF](docs/workbench/team-paidf.md) covers
  136. new-user sign-in, local groups, scoped allocations, and 20 LeRobot episodes
  137. submitted through the shared CLI or SDK.
  138. <a id="pick-your-first-win"></a>
  139. ## Choose your first workload
  140. | Result | Guide | Main requirement |
  141. | --- | --- | --- |
  142. | Generated images or video | [Cosmos 3 generation](docs/quickstart.md#standalone-cosmos-3-generation) | Compatible GPU and model access |
  143. | Augmented video with evaluation and curation reports | [PAIDF + Cosmos 3](workflows/guides/paidf-cosmos3.md) | Source video, GPU, S3, hosted inference |
  144. | A labeled video dataset | [Physical AI Data Factory](docs/workbench/guides/physical-ai-data-factory-deploy.md) | Source video, RT-core GPU, S3 |
  145. | A reconstructed 3D scene and novel views | [Neural reconstruction](docs/workbench/guides/neural-reconstruction.md) | Sensor capture and RT-core GPU |
  146. | A robot-policy training checkpoint | [Reachy 2 + LeRobot](docs/workbench/guides/reachy2-lerobot-policy.md) | Matching recorded dataset and GPU |
  147. | A Franka training and evaluation exercise | [Franka + Genesis](docs/workbench/guides/franka-pick-and-place-genesis.md) | GPU; recorded run did not solve the task |
  148. | Quadruped reinforcement learning | [Isaac Lab](docs/workbench/guides/quadruped-isaac-lab.md) | RT-core GPU |
  149. | A G1 locomotion evaluation or training path | [G1 + SONIC](docs/workbench/guides/g1-humanoid-walk-sonic.md) | Runtime-specific GPU and checkpoints |
  150. | A browser workbench and artifact viewer | [Deploy the agent](docs/agent.md) | Terraform, S3, SSH key, Token Factory key |
  151. The [guide index](docs/workbench/guides/README.md) records validation scope.
  152. [Cookbooks](docs/workbench/cookbooks/README.md) cover longer training and data
  153. pipelines. GPU names alone do not establish image compatibility; check the
  154. [image/GPU matrix](docs/workbench/image-gpu-compatibility-matrix.md) before provisioning.
  155. ## Compose it into a workflow
  156. A workflow is a YAML state graph: each `toolRef` selects a Workbench operation;
  157. outputs become inputs to later stages. SkyPilot schedules the work on Nebius.
  158. These commands validate and plan a checked-in example **locally**:
  159. ```bash
  160. npa workbench workflow validate-spec workflows/testing/cosmos3-generate.yaml
  161. npa workbench workflow plan-spec workflows/testing/cosmos3-generate.yaml --run-id demo
  162. ```
  163. Expect a valid specification and a plan containing the `generate` stage.
  164. The example's storage values are placeholders. Planning creates no workload and
  165. does not prove its inputs, credentials, image, or GPU are ready for execution.
  166. For a real run, follow the selected guide's `prepare-run`, image preflight,
  167. `submit --runtime`, and monitoring instructions with your own project and input.
  168. Collect executed actions across physical scene variations with the
  169. [physical augmentation demo](docs/workbench/guides/physical-augmentation.md):
  170. one launch command produces an offline interactive replay, comparison film,
  171. and verified action dataset from actual RTX simulation.
  172. See the [workflow catalog](workflows/README.md),
  173. [authoring guide](docs/workbench/npa-workflow-guide.md), and
  174. [run lifecycle](docs/run-lifecycle.md). The canonical
  175. [14-stage Sim2Real pipeline](docs/workbench/guides/sim2real-workflow.md) uses this
  176. same runtime at [`workflows/main/sim2real.yaml`](workflows/main/sim2real.yaml)
  177. and requires its own prepared images, task data, and resource profiles.
  178. Its manual runbook starts with a fresh CLI install and a qualified five-image
  179. release gate before provisioning; running it requires no agent deployment.
  180. Its [data contracts](docs/workbench/guides/sim2real-data-contracts.md) preserve
  181. simulator episode resets across sparse samples and exclude reset intervals from
  182. training credit. The older `sim2real/runbook.yaml` is a legacy path.
  183. Fresh stock-Franka training uses `NPA_BYO_ISAAC_BOOTSTRAP=auto` to initialize its
  184. learned actor from native Isaac demonstrations on the sealed training split,
  185. then runs the full PPO pass. Resumed checkpoints and custom robots retain their
  186. existing training path. Set `NPA_BYO_ISAAC_BOOTSTRAP=none` to disable the warm
  187. start, or use `franka-cartesian-gru-distillation-v2` to require its supported
  188. task boundary.
  189. See the [training and evaluation contract](docs/workbench/guides/sim2real-workflow.md).
  190. The canonical stock-Franka scenario task uses upstream task-space arm drives
  191. with robot gravity compensation and 200 Hz physics. Its policy still acts at
  192. 50 Hz. Training and held-out artifacts attest this same control profile.
  193. <a id="whats-in-the-box"></a>
  194. ## Find a tool or integration
  195. | Task | Reference |
  196. | --- | --- |
  197. | Discover tools by task | [Workbench docs](docs/workbench/README.md) |
  198. | Use FA4 in your own RTX PRO 6000 container | [RTX PRO 6000 FA4 adoption guide](docs/workbench/guides/rtx6000-fa4.md) |
  199. | Compare standalone FA2 and tuned FA4 | [RTX PRO 6000 measurements and actual renders](docs/workbench/fa2-fa4-validation.md) |
  200. | Find a command or option | [CLI index](docs/cli/README.md), then `npa workbench <tool> --help` |
  201. | Call a tool from Python or HTTP | [CLI / SDK walkthrough](docs/workbench/cli-sdk-yaml-walkthrough.md) — supported interfaces vary by tool |
  202. | Develop a native Ray application | [Ray guide](docs/workbench/ray.md) |
  203. | Inspect or share run outputs | [Rerun](docs/workbench/rerun-sharing.md) · [Foxglove/MCAP](docs/workbench/foxglove-export.md) |
  204. | Diagnose a failed run | [Troubleshooting](docs/workbench/troubleshooting/known-footguns.md) |
  205. ## When you're done, tear it down
  206. Cancel active jobs before deleting their infrastructure. Preview project cleanup:
  207. ```bash
  208. npa destroy --project "<alias>" --all
  209. ```
  210. This is read-only until you pass `--yes`. Review the listed resources and save
  211. needed artifacts before deleting storage. [Teardown](docs/teardown.md) covers
  212. service, controller, cluster, and storage cleanup. Local `npa cleanup` alone
  213. does not stop cloud resources.
  214. ## Container images
  215. Use the [public image catalog](docs/workbench/container-image-catalog.md) to pull
  216. available GHCR images. Repository-owned runtime defaults already select GHCR;
  217. `npa configure` needs no registry setup. Some images fetch separately licensed
  218. runtimes or model weights at startup, so check their access requirements too.
  219. For modified or private images, follow the
  220. [build and packaging guide](docs/workbench/container-packaging.md) and select
  221. the image explicitly. `NPA_REGISTRY` alone does not change runtime defaults.
  222. ## Validated on Nebius
  223. The [image/GPU matrix](docs/workbench/image-gpu-compatibility-matrix.md) records
  224. hardware tests and distinguishes them from expected compatibility. Each guide
  225. states what its run established: a working container, a generated artifact, a
  226. training checkpoint, or measured task performance. A smoke test is not evidence
  227. of policy convergence. See also the
  228. [LeRobot benchmarks](docs/workbench/cookbooks/lerobot-gpu-benchmarks.md) and
  229. [planned partner capabilities](docs/architecture/partner-skills-roadmap.md).
  230. ## Repository layout
  231. | Directory | Contents |
  232. | --- | --- |
  233. | [`npa/`](npa/README.md) | Python package, CLI, SDK, and tests |
  234. | [`workflows/`](workflows/README.md) | Declarative workflow catalog and runbooks |
  235. | [`docs/`](docs/README.md) | Setup, tool guides, cookbooks, and references |
  236. | [`deploy/cluster/`](deploy/cluster/README.md) | Managed Kubernetes Terraform wrapper |
  237. | [`skills/`](skills/index.yaml) | Operating and contribution instructions for agents |
  238. | [`workbench/mlflow/`](workbench/mlflow/README.md) | Local MLflow and Postgres stack |
  239. | [`research/`](research/lerobot-deploy/README.md) | Historical standalone deployment research |
  240. ## Documentation
  241. Use the [documentation index](docs/README.md) to find setup, operations, and
  242. contributor references. For workflow operations, start with
  243. [manual runs](docs/workbench/guides/manual-workflow-operations.md),
  244. [episode batches and capacity](docs/workbench/guides/paidf-dataset-batches.md), or
  245. [quality reports](docs/workbench/insights-reports.md).
  246. Report a broken example with its command, `npa` version,
  247. and redacted error in [GitHub Issues](https://github.com/nebius/nebius-physical-ai/issues).
  248. Keep credentials and private infrastructure identifiers out of issue text.
  249. ## Contributing
  250. For dedicated CI capacity, operators can set the repository Actions variables
  251. `NPA_CI_PRIORITY_RUNNER` and `NPA_CI_TEST_RUNNER` to approved Ubuntu runner labels.
  252. Both default to `ubuntu-latest`; neither reserves capacity by itself. See
  253. [validation concurrency](CONTRIBUTING.md#validation-concurrency) for routing and setup.
  254. The [temporary CPU runner guide](.github/ci-runners/README.md) covers disposable
  255. Nebius workers. `make ci-runners-down` restores routing and safely drains the
  256. configured pool; `make ci-runners-status` reports its state.
  257. See [CONTRIBUTING.md](CONTRIBUTING.md) for the development environment, required
  258. checks, and PR process. [The package README](npa/README.md#developing-and-testing-npa)
  259. has the shortest test commands. Update the relevant documentation and
  260. [root skill](skills/index.yaml) when changing behavior.
  261. Run `make precheck` for fast local CI checks. After committing, fetch main and run
  262. `make merge-precheck` to check the combined dependency inputs. The
  263. [merge-readiness guide](CONTRIBUTING.md#merge-readiness-and-queue-rejections)
  264. also explains the automatic PR comments for merge-queue rejections.
  265. Security disclosures: [SECURITY.md](SECURITY.md).
  266. ## License
  267. [Apache License 2.0](LICENSE). Third-party software, models, and datasets retain
  268. their own licenses and access terms.
  269. ### MJLab robot learning
  270. [MJLab integration](docs/workbench/mjlab.md) provides native training and resume,
  271. measured evaluation, ONNX export, a scoped authenticated service, and CLI/SDK
  272. clients. Use the [train/evaluate workflow](workflows/testing/mjlab-train-eval.yaml)
  273. on a Nebius GPU with an explicitly built MJLab image. Native GPU acceptance
  274. results are recorded in the guide; public image promotion remains gated.
  275. `eval --video` publishes the rendered MP4 and a self-contained HTML report with
  276. measured episode results and checkpoint provenance.
  277. The former deterministic scoring placeholder has been removed.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.