Refactor Op
CVCUDA/CV-CUDA
Find and safely apply per-operator refactoring / redundancy-reduction opportunities in a CV-CUDA operator (near-duplicate Tensor/VarShape kernels, reinvented shared utilities, dead code).
Runs and validates an end-to-end Mission Control demo in a locally installed Isaac Sim, with a Nova Carter robot driven through a Python server.
$ npx skills add NVIDIA/skills --skill isaac-mission-control-showcase -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install NVIDIA/skills isaac-mission-control-showcase --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/NVIDIA/skills.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/isaac-mission-control-showcase .claude/skills/isaac-mission-control-showcase && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "isaac-mission-control-showcase" agent skill from https://github.com/NVIDIA/skills/tree/main/skills/isaac-mission-control-showcase into .claude/skills/isaac-mission-control-showcase/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "isaac-mission-control-showcase", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/NVIDIA/skills/tree/main/skills/isaac-mission-control-showcaseType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add NVIDIA/skills --skill isaac-mission-control-showcase -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install NVIDIA/skills isaac-mission-control-showcase --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/NVIDIA/skills.git skills-src && mkdir -p .agents/skills && cp -r skills-src/skills/isaac-mission-control-showcase .agents/skills/isaac-mission-control-showcase && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "isaac-mission-control-showcase" agent skill from https://github.com/NVIDIA/skills/tree/main/skills/isaac-mission-control-showcase into .agents/skills/isaac-mission-control-showcase/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "isaac-mission-control-showcase", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add NVIDIA/skills --skill isaac-mission-control-showcase -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install NVIDIA/skills isaac-mission-control-showcase --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/NVIDIA/skills.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/skills/isaac-mission-control-showcase .cursor/skills/isaac-mission-control-showcase && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "isaac-mission-control-showcase" agent skill from https://github.com/NVIDIA/skills/tree/main/skills/isaac-mission-control-showcase into .cursor/skills/isaac-mission-control-showcase/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "isaac-mission-control-showcase", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/NVIDIA/skills.git --path skills/isaac-mission-control-showcase--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add NVIDIA/skills --skill isaac-mission-control-showcase -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install NVIDIA/skills isaac-mission-control-showcase --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/NVIDIA/skills.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/skills/isaac-mission-control-showcase .gemini/skills/isaac-mission-control-showcase && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "isaac-mission-control-showcase" agent skill from https://github.com/NVIDIA/skills/tree/main/skills/isaac-mission-control-showcase into .gemini/skills/isaac-mission-control-showcase/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "isaac-mission-control-showcase", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install NVIDIA/skills isaac-mission-control-showcaseInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add NVIDIA/skills --skill isaac-mission-control-showcase -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/NVIDIA/skills.git skills-src && mkdir -p .github/skills && cp -r skills-src/skills/isaac-mission-control-showcase .github/skills/isaac-mission-control-showcase && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "isaac-mission-control-showcase" agent skill from https://github.com/NVIDIA/skills/tree/main/skills/isaac-mission-control-showcase into .github/skills/isaac-mission-control-showcase/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "isaac-mission-control-showcase", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add NVIDIA/skills --skill isaac-mission-control-showcase -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install NVIDIA/skills isaac-mission-control-showcase --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/NVIDIA/skills.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/skills/isaac-mission-control-showcase .opencode/skills/isaac-mission-control-showcase && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "isaac-mission-control-showcase" agent skill from https://github.com/NVIDIA/skills/tree/main/skills/isaac-mission-control-showcase into .opencode/skills/isaac-mission-control-showcase/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "isaac-mission-control-showcase", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
isaac-mission-control-showcaseRuns and validates an end-to-end Mission Control demo in a locally installed Isaac Sim, with a Nova Carter robot driven through a Python server.
A bundled runner launches the default small-warehouse Nova Carter scenario in the Isaac Sim GUI window on the host display, and every stage operation goes through the isaac-sim-remote Python server on the port set in ISAAC_PYTHON_PORT. There is no WebRTC, browser viewer or containerized Isaac Sim. Without another scenario, it uses a canonical Isaac 6.1 warehouse with a deterministic circular route.
Before running, it inspects the machine read-only to find an Isaac Sim installation, its version, isaac-sim.sh, the bundled ROS 2 bridge libraries, a usable display and any running instance, and it picks a compatible runtime without prompting when compatibility can be confirmed. Installations or version changes are handed to the isaac-sim-installation skill. It coordinates five upstream skills resolved at runtime, prepared with scripts/doctor.py dependencies --prepare.
A strict safety rule covers swap: the agent never creates, resizes, enables or removes host swap unless you approve the specific operation, path and size, and a failed memory check is reported rather than fixed that way. The folder also holds references for cloud stack bring-up, fleet composition changes and map changes, a navigation map asset and evals.
7 steps, taken from the first numbered list in SKILL.md.
Read from SKILL.md and the folder at commit dfdd080. It shows what the files ask for, not the result of running them.
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Ships 1 file in scripts/ (Python, from the files we listed), which the agent can run.
Shell commands in SKILL.md call:
bashpython3dockerFrom the folder's file list and the shell code blocks in SKILL.md.
Hosts in commands or code, which the agent is likely to contact:
omniverse-content-production.s3-us-west-2.amazonaws.comFrom URLs in SKILL.md, links to its own repository left out.
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Isaac Sim Mission Control Showcase loads about 4.8k tokens when it runs, and up to ~11k if it reads all its reference files. Until then it costs about 181 tokens; SKILL.md has 2,391 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.
The full file from NVIDIA/skills at commit dfdd080, republished under its Apache-2.0 licence (© NVIDIA). 2,391 words, ~4,801 tokens.
.claude/skills/isaac-mission-control-showcase/SKILL.md (or your agent's skills folder). This skill also uses 32 other files; get the full folder from GitHub.Use the bundled runner for the default small-warehouse Nova Carter showcase.
It owns the integration and coordinates five upstream skills that it resolves
at runtime rather than vendoring. Resolve those dependencies once with
python3 "$SKILL_DIR/scripts/doctor.py" dependencies --prepare before the first run.
Never create, resize, enable, or remove host swap without the user's explicit approval of the specific operation, device or file path, and size. For example, "Create an 8 GB swap file at /swapfile" identifies all three. A general request to run the showcase, make preflight pass, or fix a memory shortfall does not authorize any swap change. If memory preflight fails, report the shortfall and suggest freeing memory or using a larger host. Ask for the specific swap approval only if the user wants that option; do not provide swap-changing commands or perform the change before approval. Swap changes can persist after the showcase.
Before running any command, set SKILL_DIR to the absolute path of the
directory containing this loaded SKILL.md, using the skill location supplied
by the skill loader. Preserve that value in each shell invocation. Do not
derive it from the current working directory or assume a source repository
layout or a fixed installation directory.
All bundled paths in this document and its references, including scripts/,
references/, shared/, assets/, config/, and
upstream-versions.lock.json, are relative to SKILL_DIR. Read them beneath
that directory and prefix bundled script paths with "$SKILL_DIR/" when
following command examples in the references. Runtime output paths and
upstream skill paths reported by the dependency manifest retain their own
locations. Commands below can run from any working directory.
Always launch Isaac Sim from a local installation through this showcase
runner, with its GUI window on the host display. Watch the robot in that
window. Drive every stage operation through the isaac-sim-remote Python
server on ISAAC_PYTHON_PORT. There is no WebRTC, browser viewer, or
containerised Isaac Sim in this workflow.
Before preflight, inspect existing state read-only. Determine the installation
path and version, whether isaac-sim.sh and the bundled ROS 2 bridge
libraries are present, whether a display is
available, and whether an Isaac Sim is already running. Do not launch or
import Isaac Sim, start or stop a container, run an installer, or alter files.
Treat the installed version, warehouse URI, Nova Carter asset and sensor graphs, and readiness probes as one compatibility set. Mere presence of Isaac Sim is not proof of compatibility.
Apply this selection policy:
If a 6.1.0 installation is present at ISAAC_SIM_DIR, report the detected
version and continue without an unnecessary prompt.
If a different version is detected and the launcher, ROS 2 bridge, and a verified warehouse URI are all present, use it automatically — no prompt — and report the choice:
Isaac Sim
<version>detected at<path>and is usable (launcher<ready/status>, ROS 2 bridge<ready/status>). Using this installation.
Attempt the normal preflight and readiness contract; stop and report a mismatch rather than modifying that installation.
Prompt before preflight only when the version cannot be determined, or none of launcher, ROS 2 bridge, or a verified warehouse URI can be confirmed — this is the sole runtime-selection case that requires user input, because no automatic choice can be verified safe:
Isaac Sim
<version>was detected at<path>. Launcher:<ready/status>. ROS 2 bridge:<ready/status>. Choose whether to use the detected version or select or install another version.
Docker is still required for the Mission Control cloud stack and Nova Carter SIL, but never for Isaac Sim itself.
If nothing is installed, route to the resolved isaac-sim-installation
skill rather than guessing a path. Resolve it through
python3 "$SKILL_DIR/scripts/doctor.py" dependencies;
never search for it and never vendor a copy.
Never upgrade, downgrade, overwrite, delete, repair, relabel, refresh, or otherwise modify an existing Isaac Sim installation as showcase recovery. Obtain explicit user direction before stopping an already-running Isaac Sim.
If the user asks to install, upgrade, downgrade, or otherwise obtain a
different Isaac Sim version, stop the showcase workflow and hand off to the
resolved isaac-sim-installation skill. Read its runtime SKILL.md at the path
the dependency manifest reports, and follow its own gates without bypassing any
of them. Do not reproduce or restate its installation logic here. The runner
requires a local standalone installation, so preselect that method. Stop after
the installed path is reported; do not launch Isaac Sim. Resume runtime
selection only after that handoff has completed and the user asks to continue.
When that skill runs a compatibility check, prefer a generous
--timeout-seconds such as 3600; its 600-second default often expires on a
first run that is still populating shader and asset caches.
Check whether the user already specified a different map, robot platform,
simulator, or mission objective in the conversation; if not, use the
canonical default without asking. Before preflight, tell the user the exact
warehouse URI, robot ID carter01, all five waypoint coordinates in order,
route-only mission type, and mission timeout 900 seconds from
Canonical default and
Deterministic route. State that the default demo
command bash "$SKILL_DIR/scripts/run.sh" --demo uses those values without
scenario overrides. This announcement is required even if preflight later
blocks the run.
Apply Runtime selection without changing any detected installation. Resolve any user choice or installation handoff, then run the read-only preflight:
bash "$SKILL_DIR/scripts/run.sh" --preflightAsk the user to stop or relocate every reported running container that is
not explicitly permitted. Allow unrelated workloads by exact container name
with a repeated --ignore-container NAME; ignored containers still undergo
the normal port-conflict checks. Do not stop an existing Isaac Sim
automatically. Require a free ISAAC_PYTHON_PORT.
Start the stack and leave the robot idle:
bash "$SKILL_DIR/scripts/run.sh"Or run the deterministic closed route and wait for completion:
bash "$SKILL_DIR/scripts/run.sh" --demoWatch the robot in the Isaac Sim window that the runner opens. The runner loads the stage before Carter starts, so initial robot motion remains visible.
Stop only the recorded showcase resources:
bash "$SKILL_DIR/scripts/run.sh" --stopEvery run writes run-manifest.json into its work directory before anything
starts and updates it at each transition, so an interrupted run can still be
described and stopped precisely. run-result.json is the machine-readable
acceptance artifact. Inspect a run, including one that was interrupted, with:
python3 "$SKILL_DIR/scripts/showcase.py" \
run-status --work-dir <dir>A stop targets only the containers that run recorded, is idempotent, and
preserves logs and evidence. Motion evidence is persisted when observed, so a
later stationary sample never overrides a confirmed observation. Acceptance
stays strict regardless: COMPLETED mission, robot online, healthy and IDLE,
motion confirmed, no OOM, required processes healthy.
See references/troubleshooting.md for the lifecycle states, recovery procedure, motion
verdicts, Isaac Sim exit handling, host memory policy, and the ROS 2 bridge
executable check.
The runner creates a fresh directory under ${TMPDIR:-/tmp}, prints its path,
and writes all generated Mission Control configuration, maps, Isaac caches,
logs, and state there. It must not create runtime files in the caller's working
directory. If startup or demo execution fails after the Nova Carter container
is created, the runner saves its complete timestamped ROS 2 launch output to
$SHOWCASE_WORK_DIR/logs/nova-carter-ros2.log. It refreshes the same log after
stopping Nova Carter with --stop.
Two layers.
Bundled and parent-owned — everything in this package: references/,
shared/, scripts/, assets/, config/. The references/ tree holds
integration adapters and routing contracts written for this showcase. They
are not copies of upstream skills.
Runtime-resolved — the five skills the showcase orchestrates. They are owned by their own repositories, read and invoked from pinned checkouts outside this package, and never copied in.
In the table below, adapter paths are relative to the corresponding reference directory.
| Reference | Fronts | Adapter |
|---|---|---|
references/bring-up-cloud-stack/ | bring-up-cloud-stack | scripts/run.py |
references/change-map/ | change-map | scripts/run.py |
references/change-fleet-composition/ | change-fleet-composition | scripts/run.py |
references/isaac-sim-remote/ | isaac-sim-remote | scripts/run.py |
references/isaac-sim-installation/ | isaac-sim-installation | none, by design |
upstream-versions.lock.json is the single dependency declaration: the two
public GitHub repositories, their pinned refs and immutable commits, and each
skill's required entrypoints and resources.
Resolve once per host, then run:
python3 "$SKILL_DIR/scripts/doctor.py" dependencies --prepare \
--env-file "$HOME/.mission-control-showcase/state/showcase-deps.env"scripts/doctor.py runs both phases: dependencies resolves the upstream
skills and writes the manifest, host checks this machine. The dependency
phase must succeed first, because the host phase is handed the entrypoints it
resolved.
Read references/workflow.md for the stage router, and each reference's
README.md before using that stage.
The reference tells you where the upstream is; the upstream tells you how it behaves.
SKILL.md, at the path the manifest records as
skills.<name>.skill_md. Nothing in this package restates it.Do not invoke bring-up-cloud-stack with --with-sim; that starts the generic
mission-simulator, not Nova Carter SIL. The showcase owns Nova SIL, Isaac Sim
launch and shutdown, scene restore, cross-domain readiness, mission submission,
and completion evidence.
The runner performs a read-only freshness check and the adapters consume the persisted manifest. Normal execution never clones, fetches, pulls, installs, or updates anything.
With MISSION_CONTROL_SHOWCASE_REQUIRE_PREFLIGHT=1, a reference that cannot
load a ready manifest blocks rather than falling back to discovery. An adapter
exits 3 with BLOCKED [<component>]: … naming the skill, what was expected,
and the preflight command. Never work around a blocker by vendoring a skill.
The default is:
ISAAC_SIM_DIR (default ~/isaacsim),
launched with its GUI window. Isaac Sim 6.1.0 is the last-known-good runtime.
This is a default, not a compatibility ceiling.https://omniverse-content-production.s3-us-west-2.amazonaws.com/Assets/Isaac/6.1/Isaac/Environments/Simple_Warehouse/warehouse.usd.nvcr.io/nvidia/isaac/nova_carter_sil:release-3.2.carter01.8226.9001, matching the packaged Mission Control stack.77 for both Isaac Sim and Carter SIL, with normal host
transport. The isolated domain prevents unrelated domain-0 DDS participants
from exhausting Carter SIL memory while preserving Isaac-to-SIL discovery.The exact warehouse URI is passed into Isaac Sim and verified after stage
load. Never discover the warehouse through get_assets_root_path, a local
cache, a source checkout, or a filesystem search.
The runner configures Mission Control and starts the cloud stack, then launches Isaac Sim and restores the stage while the cloud initializes. Carter SIL starts last, so the Isaac Sim window shows the robot's initial motion.
For a user-selected noncanonical Isaac version that is already installed:
Require a compatible installation and a canonical asset URI.
Pass both explicitly:
ISAAC_SIM_DIR=<resolved-install-path> \
WAREHOUSE_USD_URI=<resolved-uri> \
bash "$SKILL_DIR/scripts/run.sh"Do not infer a future URI by substituting a version number. A local installation carries no inspectable version tag, so the caller owns the asset contract.
If the requested version is not already installed, stop and hand off to
isaac-sim-installation; do not install it inside this workflow.
carter01 is only the default. Set another identity with either:
bash "$SKILL_DIR/scripts/run.sh" \
--robot-name my_robot --demoor ROBOT_NAME=my_robot.
The runner applies the same value to Mission Control fleet configuration,
Mission Client serial_number, MQTT bridge client names, readiness polling,
and mission submission. Do not change only the mission payload.
The default replay is a closed route:
[
{"x": -5.1, "y": 1.4},
{"x": -2.625, "y": 1.2},
{"x": -0.2, "y": -2.075},
{"x": -2.625, "y": -5.35},
{"x": -5.1, "y": 1.4}
]Submit it as a route-only mission with timeout 900 and solver
NVIDIA_CUOPT. Do not add start/end locations or iterations.
For custom coordinates, first bring up the stack without --demo, snap the
requested points through WPG nearest_nodes, validate with
visualize_route, and submit only the resulting routable points.
The runner must not submit a mission until all of these are observed:
/clock advances;/scan LaserScan uses frame front_2d_lidar and has a timestamp
within one second of simulation time;map at that exact sensor timestamp;map -> base_link resolves;IDLE, error-free,
position-initialized, and near the reset pose.Treat an initial exact-time LiDAR TF miss as a bounded startup condition.
Continue consuming newer scan samples until one resolves into map, or fail
at the readiness deadline with the latest measured timestamp. Never replace
this with a latest-TF lookup.
On failure, report the failed observation and measured values. Do not claim a clock, TF, costmap, or permission root cause without those measurements.
The bundled carter_warehouse_navigation.png and companion YAML are the
single map resource. The runtime copies that pair into Mission Control's
generated config and Carter SIL's read-only /maps mount. Mission Control
uses the same PNG and metadata to generate the WPG resource under
carter_warehouse_navigation.png. The runner verifies the source, Mission
Control, and Carter hashes before startup, then verifies the active WPG map
identity, Carter launch option, and container-mounted bytes. Never recursively change permissions on a
source tree or apply a permission fix after services start.
--preflight: read-only installation, display, Docker, GPU, dependency,
port, image, and asset checks.--demo / --replay: startup, readiness, circular mission, canonical
mission completion, and pose-motion evidence.--stop: stop the last recorded stack and SIGTERM the Isaac Sim process,
without deleting containers, images, volumes, caches, or logs.--dry-run: print resolved commands without changing runtime state.--work-dir PATH: use a caller-selected new or empty runtime directory.--no-pull: require required images to already be local.--ignore-container NAME: allow one already-running container by exact
name during preflight and startup; repeat the option for each allowed
container. Port conflicts remain failures.Supported environment overrides are ISAAC_SIM_DIR, WAREHOUSE_USD_URI,
NOVA_CARTER_IMAGE, NOVA_CARTER_MEMORY_LIMIT,
NOVA_CARTER_MEMORY_SWAP_LIMIT, ROBOT_NAME, SHOWCASE_GPU_DEVICE,
SHOWCASE_ROS_DOMAIN_ID, SHOWCASE_ROS_LOCALHOST_ONLY,
ISAAC_PYTHON_PORT, and SHOWCASE_WORK_DIR. Keep the default isolated domain
unless it conflicts with another local ROS deployment. Apply the same domain
and localhost setting to Isaac Sim and Carter SIL. Do not enable localhost-only
transport without verifying cross-process discovery between the host Isaac Sim
bridge and host-network Carter SIL.
A successful --demo requires Mission Database to report the submitted
mission COMPLETED, the same robot to return to IDLE without errors, and
Mission Database poses to prove motion. Robot state, logs, a visible Isaac Sim
window, or a single screenshot alone are not completion.
Do not run docker compose down -v, remove containers, prune Docker state, or
delete the run directory as automatic recovery.
© NVIDIA, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
SKILL.md and 32 other files (scripts, references, assets) in skills/isaac-mission-control-showcase of NVIDIA/skills.
Open the folder on GitHubat commit dfdd080
Isaac Sim Mission Control Showcase next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Isaac Sim Mission Control Showcase this skillNVIDIA/skills | 3.5k | — | ~4.8k | Automated safety check: Pass | Apache-2.0 | |
| Refactor OpCVCUDA/CV-CUDA | 2.7k | — | ~1.5k | Automated safety check: Pass | Custom licence | |
| Megatron-LM Linting and FormattingNVIDIA/Megatron-LM | 18k | — | ~316 | Automated safety check: Pass | Apache-2.0 | |
| Tilelang Developeryzlnew/infra-skills | 149 | — | ~2.4k | Automated safety check: Pass | None | |
| Dstack Prototypingdstackai/dstack | 2.3k | — | ~1.6k | Automated safety check: Pass | MPL-2.0 | |
| Gds DiagNVIDIA/MagnumIO | 125 | — | ~1.6k | Automated safety check: Pass | Apache-2.0 |
CVCUDA/CV-CUDA
Find and safely apply per-operator refactoring / redundancy-reduction opportunities in a CV-CUDA operator (near-duplicate Tensor/VarShape kernels, reinvented shared utilities, dead code).
NVIDIA/Megatron-LM
Runs the Megatron-LM autoformat script and its linting tools before a pull request, and keeps Python imports in order with isort.
yzlnew/infra-skills
Write, optimize, and debug high-performance AI compute kernels using TileLang (a Python DSL for GPU programming).
dstackai/dstack
Use with the dstack skill for model-serving work when the image, serving command, resources, backend/fleet choice, or service behavior is not proven.
NVIDIA/MagnumIO
A skill your agent uses when diagnosing NVIDIA GPUDirect Storage with this repository: choose and run the right gds-diag.py subcommand, interpret its output, and explain operator next steps without…
Luna5ama/Alpha-Piscium
Drive NVIDIA Nsight Graphics 2026.1+ from the command line for GPU performance analysis, frame capture, frame trace inspection, draw-call inspection, NVTX/D3DPERF stage timing, replay metadata…
NVIDIA/skills
A skill your agent uses when the user wants to deploy, run, debug, tear down, or call the REST API of the RTVI-CV 2D detection / tracking microservice.
NVIDIA/skills
Generates, validates, compares and explains HOLOLINK_def.svh macro files for the HSB IP, using bundled Python scripts and asking before it writes anything.
NVIDIA/skills
Orchestrates defect image generation for PCBA, metal surface and glass inspection with NVIDIA Cosmos AnomalyGen on OSMO, from cold-start Day 0 to real-photo Day 1 labeling.
NVIDIA/skills
Orchestrates video data augmentation and auto-labeling workflows on OSMO, from flow selection and preflight checks to submission, monitoring and output download.
NVIDIA/skills
Runs NVIDIA TAO Data Services KPI analysis on object detection results, comparing predictions to ground truth and writing per-class precision, recall and AP to a CSV.
NVIDIA/skills
Runs TAO Data Services gap analysis that compares ground-truth and predicted boxes to find weak images by per-class recall, precision and AP50.
Works with
Categories
Runs and validates an end-to-end Mission Control demo in a locally installed Isaac Sim, with a Nova Carter robot driven through a Python server. A bundled runner launches the default small-warehouse Nova Carter scenario in the Isaac Sim GUI window on the host display, and every stage operation goes through the isaac-sim-remote Python server on the port set in ISAAC_PYTHON_PORT. There is no WebRTC, browser viewer or containerized Isaac Sim.
Isaac Sim Mission Control Showcase fits situations like: running a demo of Mission Control driving a simulated warehouse robot; replaying the showcase scenario to validate an Isaac Sim setup; diagnosing why the integrated small-warehouse scenario fails to run.
Run `npx skills add NVIDIA/skills --skill isaac-mission-control-showcase -a claude-code`. Or copy the skill folder (skills/isaac-mission-control-showcase in NVIDIA/skills) into .claude/skills/isaac-mission-control-showcase in your project. Claude Code loads it when a task matches its description.
Run `npx skills add NVIDIA/skills --skill isaac-mission-control-showcase -a codex`. Or copy the skill folder (skills/isaac-mission-control-showcase in NVIDIA/skills) into .agents/skills/isaac-mission-control-showcase in your project. Codex loads it when a task matches its description.
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add NVIDIA/skills --skill isaac-mission-control-showcase -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/isaac-mission-control-showcase, .gemini/skills/isaac-mission-control-showcase, .github/skills/isaac-mission-control-showcase and .opencode/skills/isaac-mission-control-showcase in your project.
Going by SKILL.md and its folder, Isaac Sim Mission Control Showcase needs Python for the scripts in its folder and the command-line tools its instructions call (bash, python3 and docker). Our summary lists: A local Isaac Sim installation with isaac-sim.sh and a display; Python 3 for the bundled runner and doctor script; The five upstream skills the runner resolves at runtime.
SKILL.md names 1 domain. In commands or code: omniverse-content-production.s3-us-west-2.amazonaws.com; the agent is likely to contact it when it follows the instructions. This is read from the text; nothing was executed.
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.
Isaac Sim Mission Control Showcase is published under the Apache-2.0 licence (declared in SKILL.md). It allows redistribution, so the full SKILL.md is shown on this page.
About 4.8k tokens (SKILL.md is roughly 19k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 6.6k tokens, read only when the agent opens those files.
Skills that share tags, products or a category with Isaac Sim Mission Control Showcase: Refactor Op (CVCUDA/CV-CUDA, 2.7k stars), Megatron-LM Linting and Formatting (NVIDIA/Megatron-LM, 18k stars), Tilelang Developer (yzlnew/infra-skills, 149 stars) and Dstack Prototyping (dstackai/dstack, 2.3k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
NVIDIA (a GitHub organization, an official publisher) maintains it in NVIDIA/skills, which has 3,546 GitHub stars. The repository holds 386 skills in this directory. The repository was last updated on October 9, 2026.
Source: NVIDIA/skills on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.