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 agentmods add skills/zimoliao/scholaraio/lammpsnpx skills add ZimoLiao/scholaraio --skill lammpsgit clone --depth 1 https://github.com/ZimoLiao/scholaraioWhat 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 | $0.00040 | $0.02547 |
| Opus 5 | $0.00020 | $0.01273 |
| Sonnet 5 | $0.00008 | $0.00509 |
| Haiku 4.5 | $0.00004 | $0.00255 |
Grade A, and why
lammps 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 yesterday.
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 — 200 lines — stays where its author put it; the contents beside it link to each section on GitHub.
LAMMPS 材料科学模拟
用 LAMMPS 做材料科学分子动力学模拟：晶体构建、势函数选择、形变/冲击、结构分析、可视化。
本 skill 故意保持轻量：
- 它负责告诉 agent 什么时候该用 LAMMPS、该遵守什么科学规范、完整工作流长什么样
- 它不承担完整接口手册的职责
- 具体命令、参数、语法、package 限制，统一去查
scholaraio toolref
Agent 默认协议（toolref-first）
对 LAMMPS 问题，agent 默认按这个顺序工作：
- 先判断问题属于哪类对象：
pair_style、fix、compute、dump、region、boundary、run流程 - 写输入脚本前，优先查高风险命令页，而不是凭记忆拼装
- 命令名和用户说法不一致时，优先用
search找主入口，再用show - 如果
toolref已能回答，就不要在 skill 里重复写手册 - 如果
toolref命中不好或某个 package 页面缺失，agent 应先完成任务，再把它标记为维护层缺口，而不是让用户自己补
这意味着：
- 用户不该自己去打磨
fix/pair_style的映射关系 - agent 应自己消化
fix npt -> fix_nh、pair style eam -> pair_eam这类入口差异 - 只有反复出现的缺口才进入正式 onboarding
前置条件
# 安装（含 GPU 支持）
conda install -c conda-forge lammps
# 可视化
pip install ovito
验证：lmp -h 应显示已安装的 packages（需包含 GPU、MANYBODY、EXTRA-COMPUTE）。
GPU 加速：package gpu 4 在输入脚本开头启用，suffix gpu 自动为支持的 pair_style 加 /gpu 后缀。
并行运行约束：
- 启动 MPI 作业时，优先使用 LAMMPS 所在环境自带的
mpirun/mpiexec - 不要默认混用系统里的另一套 MPI launcher；如果
lmp链接的libmpi与 launcher 不一致，可能出现“进程活着但无日志输出”的假启动挂起 - 需要绑核或做 rank pinning 时，先用小规模 MPI 诊断确认启动和绑定语法，再放大到正式规模
何时使用
适合：
- 金属、陶瓷、半导体、纳米材料的经典 MD
- 力学性质、相变、位错/缺陷演化、冲击波、热输运
- 已有合适经验势函数的体系
不适合：
- 需要显式电子结构精度时，优先考虑 DFT / Quantum ESPRESSO
- 势函数没有可靠文献依据时，不要直接开算
Toolref 优先
当 agent 不确定命令、参数、限制、输出字段时，先查 toolref，再写输入脚本。
常用查法：
scholaraio toolref search lammps "nose hoover thermostat"
scholaraio toolref show lammps fix_nh
scholaraio toolref show lammps pair_eam
scholaraio toolref show lammps compute_cna_atom
scholaraio toolref show lammps fix_deform
推荐习惯：
- 写脚本前先查
pair_style/fix/compute/dump - 遇到 package 依赖时先用
toolref show看 Restrictions - 遇到模糊概念先用
toolref search，确定候选命令后再show
如果遇到覆盖缺口：
- 先用官方 LAMMPS 文档继续完成任务
- 明确说明这里是
toolref覆盖/排序缺口，不是用户操作错误 - 不要让用户为了当前任务先停下来维护文档层
- 如果官方手册没有解释清楚实际运行异常、MPI 启动问题、版本兼容或绑定语法，继续上网检索官方文档、社区讨论和已知问题，不要猜
核心流程
知识库协作模式
- 用
scholaraio usearch "<材料/现象>"检索相关论文 - 从论文提取：势函数选择、晶格常数、实验基准值（相变压力、弹性常数等）
- 在输入脚本注释中标注参数来源
- 计算完成后与文献数据定量对比
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.
- yesterday First seen · 200 lines · 40 tokens per session scan A baa4d7911490
lammps is a skill published in the GitHub repository ZimoLiao/scholaraio (566 stars, last pushed yesterday), licensed MIT. It adds 40 tokens to every session and 2,547 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 2026-08-30.
Other skills, from other repositories
create-lab
Create a simulator-first organization laboratory scaffold with policy and validation. Use when bootstrapping a separate lab repository from OpenSDL.
arxiv
Skill "arxiv" from 0xmariowu/Autosearch, covering skill attribution, when to choose it, how to search (planned), known quirks and quality bar.
dblp
Skill "dblp" from 0xmariowu/Autosearch, covering skill attribution, when to choose it, how to search, known quirks and quality bar.
papers
Skill "papers" from 0xmariowu/Autosearch, covering skill attribution, known quirks and quality bar.
wikipedia
Authoritative encyclopedia articles via the Wikipedia Action API (English edition).
pubmed
Use for biomedical and life-science literature searches when query expects PubMed articles, clinical trials, or medical research papers.