Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/D-Robotics/mossnpx agentmods add skills/d-robotics/moss/rdk-llm-deploymentWrote 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.
[](https://agentmods.dev/skills/d-robotics/moss/rdk-llm-deployment)<a href="https://agentmods.dev/skills/d-robotics/moss/rdk-llm-deployment"><img src="https://agentmods.dev/badge/skills/d-robotics/moss/rdk-llm-deployment.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 6 findings, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Tool Misuse · line 88 Tool calls are chained to bypass individual safety checks or escalate capabilities beyond what any single tool call would allow.Fix: Limit tool chaining depth and validate the output of each tool before passing it to the next. Require explicit user approval for multi-step chains.
- medium Privilege Escalation · line 88 Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.Fix: Avoid sudo/root unless strictly required. Prefer least-privilege patterns. If elevation is needed, document the justification and scope.
- medium Privilege Escalation · line 89 Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.Fix: Avoid sudo/root unless strictly required. Prefer least-privilege patterns. If elevation is needed, document the justification and scope.
- medium Privilege Escalation · line 91 Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.Fix: Avoid sudo/root unless strictly required. Prefer least-privilege patterns. If elevation is needed, document the justification and scope.
- medium Privilege Escalation · line 106 Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.Fix: Avoid sudo/root unless strictly required. Prefer least-privilege patterns. If elevation is needed, document the justification and scope.
- medium Privilege Escalation · line 119 Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.Fix: Avoid sudo/root unless strictly required. Prefer least-privilege patterns. If elevation is needed, document the justification and scope.
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 | $0.00341 | $0.04916 |
| Opus 5 | $0.00170 | $0.02458 |
| Sonnet 5 | $0.00068 | $0.00983 |
| Haiku 4.5 | $0.00034 | $0.00492 |
Grade B, and why
rdk-llm-deployment scanned grade B with 2 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 8d 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.
Asks for rootmediumPrivilege escalation
A mod that escalates privileges can change anything on the machine, not only the project.
sudo apt update && sudo apt install -y hobot-dnn # update on-board dnn Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
wget https://d-robotics-aitoolchain.oss-cn-beijing.aliyuncs.com/llm_s600/1.0.2/D-Robotics_LLM_S600_1.0.2_SDK.tar.gz How it starts
The opening of the file, as written. The whole thing — 164 lines — stays where its author put it; the contents beside it link to each section on GitHub.
RDK On-Device LLM / VLM / Voice Deployment
Get a chatbot, a vision-language model, or a full voice assistant running on the board (not in the cloud). The single most important fact: the runtime stack is chosen by the board, and the two stacks do not overlap — hobot_llamacpp is the X5/S100 path, and S600 uses a completely different runtime (oellm_runtime from the D-Robotics_LLM_S600 SDK). Pick the stack from the board first; everything else follows.
Sources: official D-Robotics repos hobot_llamacpp, hobot_llm, sensevoice_ros2, hobot_tts, xiaozhi-in-rdk, oellm_server; plus rdk_s_doc
LLM_Toolchain(S100/S600) and the rdk_dochobot_llm(Bloom) page. Facts verified against these on 2026-06; model lists track the repos / HuggingFace at that time.
Board → stack cheat-sheet (decide this first)
| Board | On-device LLM/VLM stack | Runtime | Artifact / model format | Notes |
|---|---|---|---|---|
| RDK X5 | hobot_llamacpp |
llama.cpp (tag b4749) + BPU |
GGUF (-GGUF-BPU) + ViT encoder .bin |
1–2B VLM fluent; Ubuntu 22.04 + Humble |
| RDK S100 / S100P | hobot_llamacpp or oellm_runtime (D-Robotics_LLM_S100 SDK) |
llama.cpp + BPU / libxlm.so |
GGUF + ViT encoder .hbm, or .hbm (march nash-e S100 / nash-m S100P) |
bigger RAM → up to InternVL3-8B; Ubuntu 22.04 + Humble |
| RDK S600 | oellm_runtime ONLY (D-Robotics_LLM_S600 SDK) |
libxlm.so (OE-LLM / LeapLLM) |
.hbm (march nash-p) |
NOT hobot_llamacpp — no S600 build flag; Ubuntu 24.04 + Jazzy |
| RDK X3 (4GB) | hobot_llm (legacy, apt) |
hobot-dnn | Bloom 1.4B tar from archive.d-robotics.cc |
4GB RAM only; Ubuntu 20.04/22.04 |
| RDK Ultra / X3 (2GB) | — | — | — | No first-party on-device LLM path |
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 8d ago First seen · 164 lines · 341 tokens per session scan B cc362e50100a
rdk-llm-deployment is a skill published in the GitHub repository D-Robotics/moss (142 stars, last pushed 12d ago), licensed MIT. It adds 341 tokens to every session and 4,916 once invoked, about $0.0017 per session on Opus 5. A static security scan graded it B with 2 findings (asks for root, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
spark-environment-setup
Set up a working ML training/inference environment on NVIDIA DGX Spark (GB10, aarch64, CUDA 13). Use when installing PyTorch/Unsloth/TRL/vLLM on DGX Spark, hitting libcudart or wheel-ABI errors on aarch64, or choosing between NGC containers and bare pip installs.
spark-memory-thermal-ops
Manage unified memory and thermals during long-running ML jobs on NVIDIA DGX Spark. Use when planning memory headroom for a training run on GB10, when a job OOMs on unified memory, or when monitoring temperature and power during multi-hour training.
spark-training-gotchas
Preflight and diagnose the ten known failure modes for ML training on NVIDIA DGX Spark. Use when a training run on DGX Spark fails to start, OOMs below the 128GB limit, slows down mid-run, or before any multi-hour training job on GB10.
llama-cpp
Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1/M2/M3 Macs, AMD/Intel GPUs, or when CUDA is unavailable. Supports GGUF quantization (1.5-8 bit) for reduced memory and 4-10× speedup vs PyTorch on CPU.
amc-run-rtsp-calibration
Calibrate a new dataset from live RTSP camera streams via the AutoMagicCalib REST API. Use when the user provides RTSP URLs or asks to calibrate live cameras; VIOS records clips, AMC ingests them, then runs calibration.
amc-run-video-calibration
Calibrates pre-recorded cam.mp4 datasets through the AutoMagicCalib REST API. Use for user-supplied local MP4s; route live RTSP streams to amc-run-rtsp-calibration.