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.
npx skills add HorizonRobotics/OE-Skills --skill j6-hbdk-compilegit clone --depth 1 https://github.com/HorizonRobotics/OE-SkillsWrote 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/horizonrobotics/oe-skills/j6-hbdk-compile)<a href="https://agentmods.dev/skills/horizonrobotics/oe-skills/j6-hbdk-compile"><img src="https://agentmods.dev/badge/skills/horizonrobotics/oe-skills/j6-hbdk-compile/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.
<a href="https://agentmods.dev/skills/horizonrobotics/oe-skills/j6-hbdk-compile"><img src="https://agentmods.dev/badge/skills/horizonrobotics/oe-skills/j6-hbdk-compile.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00097 | $0.06800 |
| Opus 5 | $0.00048 | $0.03400 |
| Sonnet 5 | $0.00019 | $0.01360 |
| Haiku 4.5 | $0.00010 | $0.00680 |
Grade A, and why
j6-hbdk-compile 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 12d 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.
How it starts
The opening of the file, as written. The whole thing — 567 lines — stays where its author put it; the contents beside it link to each section on GitHub.
J6 HBDK Compile - 通用模型编译 Skill
版本: 1.1.8
概述
本 skill 基于 YAML 配置文件驱动模型编译流程：
- 解析用户指令 -> 生成配置文件
- 用户确认配置文件 -> 是否需要修改
- 加载配置并执行编译 -> 检查 CPU 算子 -> 校验产物 -> hbm_perf -> 生成报告
交互原则（Step 2 为默认强制门禁）：
- 默认：Step 1 生成或写出
compile_config_*.yaml后，必须先执行 Step 2，向用户展示配置路径与关键字段摘要，并结束本轮回复、等待用户下一条消息。在用户给出明确肯定（如「直接编译」「确认」「可以」）之前，禁止运行compile_model.py -c或任何等价编译命令。 - 唯一例外：用户在同一条用户消息中已写明可跳过确认（如「直接编译」「跳过确认」「按上述配置立刻编译」「生成配置后马上编」），才允许在同一轮助手流程中连续执行 Step 1→3。
- 仍须单独确认（不得默认继续）：缺少模型路径；pyramid/resizer 的 mean/std 等关键项与用户描述不符或需用户拍板；检测到 CPU 算子后是否继续；编译失败后的重试或改配策略。
- 模型合法性校验失败：立即终止本 skill 流程；禁止修改
compile_model.py或用旁路脚本绕过检查（详见下文「模型合法性检查」）。
默认配置：
- 目标平台:
nash-e - 输出目录: 模型同级目录下的
compile_<timestamp>文件夹（带时间戳避免覆盖） - 删除 QDQ 节点: 通过
remove_node_type: ["Quantize"]（remove_all_qdq已废弃） - debug:
true
模型合法性检查（加载阶段）
compile_model.py 在首次加载模型时（含 --generate-config 拉取 IO 与正式编译的 load_model）会执行以下检查，不通过则立即终止（异常或非零退出）并输出明确错误信息；不得在同一轮任务里继续执行后续编译步骤或假装已加载成功。
| 检查项 | 行为 |
|---|---|
| 文件后缀 | 仅允许 .onnx、.bc；否则报错「不支持的模型类型」并停止。 |
.bc 能否 load |
若 hbdk4.compiler.load 失败，提示生成该 .bc 的 hbdk4 版本可能与当前环境不兼容，并停止。 |
.bc 阶段 |
在 load 成功后检查 inner module 的 named attributes：若存在 hbdk.target，则视为已 convert 的 quantized.bc，本流程需要 qat.bc；否则继续。判定规则与业务侧 check_current_phase 一致：无 hbdk.target 时按 qat.bc 处理。 |
.onnx |
onnx.load 成功后，图中需至少包含一个 op_type == "HzCalibration" 的节点（Horizon PTQ 标志）；否则报错并提示用户检查是否为 horizon_plugin_pytorch 导出的 PTQ ONNX。 |
硬门禁（校验失败 = 停止本 skill 工作流）
任一检查失败时，你必须立刻停止：不得继续跑 -c、不得在未换模型的情况下反复「试编译」、不得替用户编造已通过校验的结论。应原样引用终端/日志中的错误信息，并明确告知用户需在模型或环境侧完成的动作（例如改回 qat.bc、换用带 HzCalibration 的 PTQ ONNX、对齐 hbdk4 与导出工具链版本）。
对 Agent 的禁止事项（严禁绕过校验）
- 禁止为通过检查而修改
compile_model.py（包括注释、删除或改写assert_supported_model_suffix、load_bc_model_validated、assert_bc_is_qat_phase、assert_onnx_has_hz_calibration及调用它们的代码路径）。 - 禁止编写或使用其它脚本 / 临时 Python，在不做同等语义校验的情况下加载模型，仅为生成 IO、写 yaml 或推进编译。
- 禁止向用户建议通过删断言、
sed去掉检查、环境变量关闭校验等方式绕过。 - 唯一合规做法：用户更换为满足上表的模型或环境后，再重新执行本仓库中的
compile_model.py（含 Docker 场景下同步到容器内的同一份脚本，亦不得删减校验）。
What ships with it
6 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.
- 12d ago First seen · 567 lines · 97 tokens per session scan A 7b6471699b75
j6-hbdk-compile is a skill published in the GitHub repository HorizonRobotics/OE-Skills (19 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 97 tokens to every session and 6,800 once invoked, about $0.0005 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 2026-08-30.
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.
minicpm5-deploy-vllm-ascend
Deploy MiniCPM5-2B with vLLM on Huawei Ascend NPU using vLLM-Ascend. Use when the user mentions vLLM-Ascend, Ascend NPU, Huawei Ascend, CANN, torchnpu, davinci devices, or wants an OpenAI-compatible MiniCPM5 server on Ascend hardware.
minicpm5-deploy-litert
Run MiniCPM5-2B or MiniCPM5-1B on-device with Google's LiteRT-LM runtime — the litert-lm CLI or its OpenAI-compatible server on a desktop, the Kotlin API or the AI Edge Gallery app on Android, the same .litertlm bundle on CPU or GPU. Use when the user says "LiteRT", "LiteRT-LM", "litertlm", ".litertlm", "Android"…