What is add-workbench-tool?

A procedure for adding a workbench tool across its implementation, command-line interface, software library, catalog, container, tests, documentation, and skill files.

Use it when introducing a new CPU tool, containerized service, or GPU-based workbench stage from start to finish.

It prevents partially added tools that build successfully but cannot be called, packaged, tested, or deployed correctly.

add-workbench-tool — Skill for Claude Code and Codex by nebius

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent

npx skills add nebius/nebius-physical-ai --skill add-workbench-tool

Clone the repo

git clone --depth 1 https://github.com/nebius/nebius-physical-ai

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/nebius/nebius-physical-ai/add-workbench-tool/github.svg)](https://agentmods.dev/skills/nebius/nebius-physical-ai/add-workbench-tool)

Your own site

<a href="https://agentmods.dev/skills/nebius/nebius-physical-ai/add-workbench-tool"><img src="https://agentmods.dev/badge/skills/nebius/nebius-physical-ai/add-workbench-tool/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/nebius/nebius-physical-ai/add-workbench-tool"><img src="https://agentmods.dev/badge/skills/nebius/nebius-physical-ai/add-workbench-tool.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 49 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 2,604 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits

NVIDIA SkillSpector pass 7 Sept 2026

How audits are shown

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00049	$0.02604
Opus 5	$0.00024	$0.01302
Sonnet 5	$0.00010	$0.00521
Haiku 4.5	$0.00005	$0.00260

Measured 4d ago against content hash 415cd77c3c6f, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

add-workbench-tool scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/workflows/add-workbench-tool/SKILL.md · 198 lines

How it starts

The opening of the file, as written. The whole thing — 198 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Add A Workbench Tool

Adding a tool is not one edit. A minimal tool touches ~18 files across six phases, and ~12 separate gates fail if you skip one. Work the phases in order: each phase's gate can only pass once the previous phase exists.

CONTRIBUTING.md is the prose rationale for this layer and docs/architecture/contributor-context.md the architecture. This skill is the ordered execution path.

Decide The Archetype First

The archetype decides which phases apply. Getting this wrong means building a container nothing routes to, or a GPU stage that lands on the default pod.

Archetype	Runs in	Phases	Examples
In-process / CPU	default npa workflow pod	A, B, D, E	`insights`, `dataset`, `scenario_gen`
Containerized service	own image, deployed to `workbench` namespace	A–E	`detection_training`, `lancedb`
Containerized GPU stage	own image, SkyPilot task pod	A–E + image routing	`cosmos3`, `nurec`

In-process tools need no Dockerfile and no image-routing entry. Do not add a container "for symmetry" — a container that no toolRef routes to is dead weight the packaging and security gates still police.

The One Architectural Rule

One implementation, three thin clients. Put behavior in npa/src/npa/workbench/<tool_snake>/; the FastAPI service, the CLI, and the SDK all call into it. Never implement training, inference, ingest, or status logic twice.

The cleanest reference to copy is detection-training:

npa/src/npa/workbench/detection_training/service.py
npa/src/npa/workbench/detection_training/schemas.py
npa/src/npa/cli/workbench/detection_training.py
npa/src/npa/sdk/workbench/detection_training.py

LeRobot (npa/src/npa/cli/workbench/lerobot.py) is the richest tool but keeps much of its logic as CLI orchestration. Read it for CLI surface breadth, not as the layering model to copy.

Naming

Pick the name once; it derives everything else. <tool> is kebab-case, <tool_snake> is snake_case.

Read the full file on GitHub · 198 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

4d ago Changed 415cd77c3c6f
8d ago First seen · 198 lines · 49 tokens per session scan A 82add7707c2b

Subscribe to this mod's changes

add-workbench-tool is a skill published in the GitHub repository nebius/nebius-physical-ai (29 stars, last pushed today), licensed Apache-2.0. It adds 49 tokens to every session and 2,604 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.

Backend and APIs Cloud DevOps and cloud

stripe-projects

Provision SaaS services + sync creds via Stripe Projects.

NousResearch/hermes-agent · 15 tokens

azure-eventhub-dotnet

Azure Event Hubs SDK for .NET. Use for high-throughput event streaming: sending events (EventHubProducerClient, EventHubBufferedProducerClient), receiving events (EventProcessorClient with checkpointing), partition management, and real-time data ingestion. Triggers: "Event Hubs", "event streaming"…

microsoft/skills · 94 tokens

azure-mgmt-botservice-dotnet

Azure Resource Manager SDK for Bot Service in .NET. Management plane operations for creating and managing Azure Bot resources, channels (Teams, DirectLine, Slack), and connection settings. Triggers: "Bot Service", "BotResource", "Azure Bot", "DirectLine channel", "Teams channel", "bot management .NET", "create bot".

microsoft/skills · 78 tokens

cve

Look up CVE records via x cve — cached, zero-API-key, daily xz TSV. Load for cve, vulnerability id, kev, epss, nvd, cvelist, or security advisory.

x-cmd/x-cmd · 49 tokens

wikipedia

Search and read Wikipedia via x wkp — MediaWiki API, no API key, zero install; query, extract, suggest, and DDG route in one module. Load for wiki, wikipedia, encyclopedia lookup, article summary.