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/HorizonRobotics/OE-Skillsnpx agentmods add skills/horizonrobotics/oe-skills/llmcompression-operationsWrote 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/llmcompression-operations)<a href="https://agentmods.dev/skills/horizonrobotics/oe-skills/llmcompression-operations"><img src="https://agentmods.dev/badge/skills/horizonrobotics/oe-skills/llmcompression-operations/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/llmcompression-operations"><img src="https://agentmods.dev/badge/skills/horizonrobotics/oe-skills/llmcompression-operations.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 | $0.00145 | $0.03228 |
| Opus 5 | $0.00072 | $0.01614 |
| Sonnet 5 | $0.00029 | $0.00646 |
| Haiku 4.5 | $0.00015 | $0.00323 |
Grade A, and why
llmcompression-operations 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 10d 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 — 221 lines — stays where its author put it; the contents beside it link to each section on GitHub.
LLM Compression 日常操作
llm_compression 是独立于 OE / OE-LLM 的 LLM 工具链包，提供校准、评测、编译、板端推理等标准脚本。
与 llmcompression-add-model 的区分：新增模型架构支持（编写 blocks/、model.py 等）请路由到
llmcompression-add-model；运行校准/评测/编译/板端推理等日常操作使用本 Skill。
⛔ 关键规则
⛔⛔ 执行目录：必须从项目根目录运行（禁止 cd 进 llm_compression/）
这是最常见的致命错误。
cd进llm_compression/会导致 Python 循环引用，所有脚本在 import 阶段即崩溃。
llm_compression/ 目录下有一个 datasets/ 子包，与第三方库 datasets（HuggingFace）同名。当 CWD 为 llm_compression/ 时，Python 的 sys.path[0]=''（CWD）会优先搜索到本地 datasets/ 子包，遮蔽第三方库，触发：
ImportError: cannot import name 'load_dataset' from partially initialized module 'datasets'
(most likely due to a circular import)
正确做法 — 从 OE LLM 包根目录（llm_compression/ 的父目录）执行脚本：
# ✅ 正确：CWD = 项目根目录
cd ${OE_LLM_DIR} # 即 /open_explorer_llm
bash llm_compression/scripts/calib.sh --config_path llm_compression/configs/qwen3_vl.yml
# ❌ 错误：CWD = llm_compression/ → circular import
cd ${OE_LLM_DIR}/llm_compression
bash scripts/calib.sh --config_path configs/qwen3_vl.yml
所有标准脚本的路径和配置路径都必须从项目根目录相对书写：
| 操作 | ✅ 正确命令（CWD = 项目根目录） |
|---|---|
| 校准 | bash llm_compression/scripts/calib.sh --config_path llm_compression/configs/<model>.yml |
| GPU 精度评测 | bash llm_compression/scripts/torch_eval.sh --config_path llm_compression/configs/<model>.yml |
| HBM 编译 | bash llm_compression/scripts/compile.sh --config_path llm_compression/configs/<model>.yml |
| 板端精度评测 | bash llm_compression/scripts/hbm_rpc_eval.sh --config_path llm_compression/configs/<model>.yml |
| 量化误差分析 | bash llm_compression/scripts/quant_analysis.sh --config_path llm_compression/configs/<model>.yml |
⛔ 必须通过 scripts/.sh 入口执行（禁止直接调用 tools/.py）
即使你已经理解了 shell 脚本的内部实现，也禁止绕过它。 标准脚本会自动处理 PYTHONPATH、环境变量、CWD 和日志，直接调用 Python 可能触发不可预见的问题（如循环引用、路径错误）。
| 操作 | ✅ 正确（通过 shell 入口） | ❌ 禁止（直接调 Python） |
|---|---|---|
| 校准 | bash llm_compression/scripts/calib.sh --config_path <yml> |
python3 llm_compression/tools/calib.py ... |
| GPU 评测 | bash llm_compression/scripts/torch_eval.sh --config_path <yml> |
python3 llm_compression/tools/torch_eval.py ... |
| 编译 | bash llm_compression/scripts/compile.sh --config_path <yml> |
python3 llm_compression/tools/compile.py ... |
| 板端评测 | bash llm_compression/scripts/hbm_rpc_eval.sh --config_path <yml> |
python3 llm_compression/tools/hbm_rpc_eval.py ... |
| 量化分析 | bash llm_compression/scripts/quant_analysis.sh --config_path <yml> |
python3 llm_compression/tools/quant_analysis.py ... |
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.
- 10d ago First seen · 221 lines · 145 tokens per session scan A 28e495fdd4f0
llmcompression-operations is a skill published in the GitHub repository HorizonRobotics/OE-Skills (19 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 145 tokens to every session and 3,228 once invoked, about $0.0007 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.
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.