AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Molecular Properties Predictor

skill-internscience-chemclaw-molecular-properties-predictor · by InternScience

预测小分子多种物化性质(沸点、折射率、密度、黏度、表面张力等),当前已真实接入 bamboo_mixer 单分子物性模型后端。

No reviews yet
0 installs
22 views
0.0% view→install

Install

$ agentstack add skill-internscience-chemclaw-molecular-properties-predictor

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-internscience-chemclaw-molecular-properties-predictor)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
5mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Molecular Properties Predictor? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Molecular Properties Predictor

功能概述

该 skill 用于预测小分子多种物化性质

当前版本通过 Bamboo-Mixer 单分子模型返回以下 11 个性质:

  • Tm:熔点(K)
  • bp:沸点(K)
  • nD:折射率(无单位)
  • nD_liquid:液体折射率(无单位)
  • dc:介电常数(无单位)
  • ST:表面张力(mN/m)
  • density:密度(g/cm^3)
  • vis:黏度(cP)
  • vapP:蒸气压(Pa)

适用场景

当用户有如下需求时适合调用:

  • 查询一个分子的多种物化性质
  • 一次性获取熔点、沸点、密度、黏度等多个性质
  • 为下游分子筛选提供多指标输入
  • 作为拆分单项物性 skill 之前的总入口

输入形式

单分子输入

  • smiles:必填
  • name:可选
  • temperature:可选,默认 25.0

批量输入

支持 JSON 列表输入,每项至少包含:

  • smiles
  • name(可选)

输出字段

每个结果条目通常包含:

  • name
  • smiles
  • canonical_smiles
  • status
  • temperature_celsius
  • backend_used
  • model_source
  • properties
  • property_units
  • raw_backend_output

⚠️ pKa 预测特别说明

本 skill 虽然输出 pka_apka_b 字段,但不推荐用于 pKa 预测。

推荐方案: 请使用专门的 [pka-predictor](../pka-predictor/SKILL.md) skill 进行 pKa 预测。

原因: | 对比项 | molecular-properties-predictor | pka-predictor | |--------|-------------------------------|---------------| | pKa 准确度 | 中等(误差 ~0.4 单位) | 高(误差 ~0.17 单位) | | 后端 | Bamboo-Mixer 多任务模型 | Uni-pKa 专用模型 | | 微观态处理 | 无 | 支持微观态枚举 + 自由能计算 | | 输出详细度 | 仅返回数值 | 电荷态、去质子化方向、置信度等 |

示例:

# ❌ 不推荐:用本 skill 预测 pKa
python scripts/main_script.py --smiles "CC(=O)O" --name "乙酸"

# ✅ 推荐:用 pka-predictor 预测 pKa
cd ../pka-predictor && ./run_with_venv.sh --smiles "CC(=O)O" --name "乙酸" --backend unipka

⚠️ 表面张力预测特别说明

本 skill 可预测表面张力 (ST),但对于单一表面张力预测需求,推荐使用专门的 [surface-tension-predictor](../surface-tension-predictor/SKILL.md) skill。

推荐策略:

| 需求场景 | 推荐 Skill/后端 | |----------|-----------------| | 仅预测表面张力 | surface-tension-predictor | | 表面张力 + 多种物性 | molecular-properties-predictor (Bamboo-Mixer) | | 小分子 (<10 重原子) | surface-tension-predictor (baseline) | | 大分子 (≥10 重原子) | surface-tension-predictor (public_joblib) |

原因: | 对比项 | molecular-properties-predictor | surface-tension-predictor | |--------|-------------------------------|---------------------------| | 表面张力准确度 | 好(误差 ~5%) | 好(误差 ~5% baseline) | | 后端 | Bamboo-Mixer 多任务模型 | baseline / publicjoblib | | 特征数 | 隐式描述符 | 8 个 (baseline) / 130 个 (publicjoblib) | | 灵活性 | 固定 11 个性质 | 可切换后端,针对表面张力优化 | | 适用场景 | 多种物性一次性预测 | 单一表面张力预测 |

测试对比(苯甲酸): | Skill/后端 | 预测值 | 文献值 | 偏差 | |------------|--------|--------|------| | surface-tension-predictor (baseline) | 41.63 mN/m | ~44 mN/m | -5.4% ✅ | | molecular-properties-predictor (Bamboo) | 41.38 mN/m | ~44 mN/m | -5.9% ✅ | | surface-tension-predictor (public_joblib) | 33.19 mN/m | ~44 mN/m | -24.6% ⚠️ |

示例:

# ✅ 推荐:仅预测表面张力,用 surface-tension-predictor
cd ../surface-tension-predictor
python scripts/main_script.py --smiles "O=C(O)c1ccccc1" --name "苯甲酸" --backend baseline

# ✅ 推荐:同时预测多种物性,用本 skill
cd ../molecular-properties-predictor
python scripts/main_script.py --smiles "O=C(O)c1ccccc1" --properties bp,ST,density

后端说明

bamboo_mixer

定位

这是当前 skill 的真实模型后端,用于对接 Bamboo-Mixer 单分子物性模型。

当前已完成的真实接入链路

当前版本已完成以下流程:

  1. 准备单分子输入 JSON
  2. 调用 Bamboo-Mixer 的:
  • scripts/prepare_data/prepare_data.py --data_type mono
  1. 调用 Bamboo-Mixer 的:
  • scripts/test_results/mono.py
  1. 使用 mono checkpoint:
  • hf_bamboo_mixer/ckpts/mono/optimal.pt
  1. 解析输出 output_mono.json
  2. 返回 11 个物性结果

当前使用的环境变量

  • BAMBOO_MIXER_ADAPTER_PY
  • BAMBOO_MIXER_REPO
  • BAMBOO_MIXER_PYTHON
  • BAMBOO_MIXER_MONO_CKPT

推荐写法:

export BAMBOO_MIXER_ADAPTER_PY=scripts/adapters/bamboo_mixer_properties_adapter.py
export BAMBOO_MIXER_REPO=assets/bamboo_mixer
export BAMBOO_MIXER_PYTHON=assets/bamboo_mixer/.venv/bin/python
export BAMBOO_MIXER_MONO_CKPT=assets/bamboo_mixer/hf_bamboo_mixer/ckpts/mono/optimal.pt

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.