← Back to Skills Marketplace
flowertusi

Genos基因序列分析

by flowertusi · GitHub ↗ · v0.1.1 · MIT-0
cross-platform ⚠ suspicious
114
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install genos-dna
Description
使用 Genos 模型进行 DNA 序列分析。当用户提到 DNA、基因、基因组、碱基序列、ACGT 等生物信息学相关问题时使用此技能。
README (SKILL.md)

Genos DNA 序列分析技能

概述

此技能使用之江实验室的 Genos-1.2B 模型进行 DNA 序列分析。该模型是人类基因组基础模型,专门用于分析 DNA 碱基序列。

使用场景

  • DNA 序列分析
  • 基因组预测
  • 碱基频率统计
  • 序列模式识别
  • 基因组学研究相关问题

模型信息

  • 模型: Genos-1.2B
  • 参数: 12 亿
  • 架构: MoE (Mixture of Experts)
  • 词汇量: 128 (A, C, G, T, N + 特殊标记)
  • 上下文长度: 最长 1M 碱基对

使用方法

1. DNA 碱基序列分析

当用户提供 DNA 序列(如 ACGTACGT...)时,调用 analyze_dna_sequence 函数进行分析。

2. 预测下一个碱基

调用 predict_next_base 函数预测 DNA 序列中下一个可能的碱基。

3. 序列特征提取

调用 extract_sequence_features 函数提取序列的特征信息。

示例

输入格式

用户可能提供:

  • DNA 序列: ACGTACGTACGT...
  • FASTA 格式的基因序列
  • 简单描述: "帮我分析这段 DNA 序列"

输出格式

返回分析结果,包括:

  • 碱基组成统计
  • 序列长度
  • 预测结果

注意事项

  1. 此模型仅支持 DNA 碱基字符(A, C, G, T, N)
  2. 不支持中文或英文自然语言输入
  3. 输入前需去除空格、换行等非碱基字符
  4. 模型主要用于基因组学研究,不适用于对话任务

模型状态检查与启动

重要:在调用技能前,必须先检查模型是否已启动!

检查模型状态

模型状态记录在 ./scripts/.model_loaded 文件中(相对于项目根目录)。

  • 如果文件存在且内容为 loaded,表示模型已启动
  • 如果文件不存在或内容不是 loaded,需要先启动模型

启动模型

如果模型未启动,执行以下命令启动:

# 设置模型路径(可选,默认为 ./models/Genos-1___2B)
export GENOS_MODEL_PATH="./models/Genos-1___2B"

# 启动模型
python3 -c "
import sys
sys.path.insert(0, './scripts')
from genos_dna import load_model
load_model()
with open('./.model_loaded', 'w') as f:
    f.write('loaded')
print('Model loaded and status saved')
"

自动化检查

AI 助手在调用技能时应自动完成以下步骤:

  1. 检查 ./.model_loaded 文件是否存在且内容为 loaded
  2. 如果模型未启动,先执行上述启动命令
  3. 确认模型启动后再调用技能函数

使用环境变量配置

你也可以通过环境变量自定义路径:

# 设置模型路径
export GENOS_MODEL_PATH="/path/to/your/model"

# 设置状态文件路径
export GENOS_STATUS_FILE="/path/to/your/state/.model_loaded"

# 然后运行脚本
python3 your_script.py

配置文件方式

你也可以创建 config.json 文件来配置路径:

{
    "model_path": "./models/Genos-1___2B",
    "state_file": "./scripts/.model_loaded"
}
Usage Guidance
This skill appears to be what it says: a local Genos model-based DNA analysis tool. Important cautions before installing or running it: - Model code execution risk: The loader uses transformers with trust_remote_code=True. That allows Python code bundled with a model repository to run when the model is loaded. Only use models from sources you trust and inspect their code if possible. - Network downloads: install.sh can download large model files from Hugging Face, ModelScope, or GitHub. Verify the model origin and checksums; avoid supplying tokens in untrusted environments. - Run in isolation: install and run inside a container, VM, or dedicated virtualenv to limit blast radius (file system and network isolation recommended). - Path inconsistency: SKILL.md's startup example writes './.model_loaded' but scripts expect './scripts/.model_loaded' (config uses scripts/.model_loaded). Confirm which path your deployment uses to avoid false negatives when checking model state. - Review dependencies: requirements.txt installs torch/transformers which may be large and require appropriate system resources. If you need higher assurance, ask the author for the exact model download URL and checksums, or require that the skill load only vetted model code (avoid trust_remote_code) and run an integrity check before loading.
Capability Analysis
Type: OpenClaw Skill Name: genos-dna Version: 0.1.1 The skill bundle is classified as suspicious due to high-risk execution patterns and instructions. Specifically, SKILL.md contains prompt-injection-style instructions directing the AI agent to execute shell commands (python3 -c) to initialize the environment, and scripts/genos_dna.py utilizes 'trust_remote_code=True' when loading models, which is a known vector for Remote Code Execution (RCE). While these capabilities are plausibly required for the stated DNA analysis purpose and are accompanied by security warnings in the README.md, they represent significant attack surfaces without evidence of intentional malice.
Capability Assessment
Purpose & Capability
Name/description match the provided code and files. All declared behaviours (analyze, predict, extract features) are implemented in scripts/genos_dna.py. No unrelated credentials, binaries, or config paths are requested.
Instruction Scope
SKILL.md instructs the agent to check a model-status file and to start the model if needed (by calling load_model). That is within the skill's purpose. There is a small inconsistency: SKILL.md's example startup snippet writes './.model_loaded' while the code expects './scripts/.model_loaded' (config and scripts use './scripts/.model_loaded' by default). The instructions direct running Python code that will load model code—appropriate for this skill but it means arbitrary model-provided code may run when the model is loaded.
Install Mechanism
No registry install spec, but an included install.sh optionally downloads the model from Hugging Face, ModelScope, or GitHub Releases—well-known hosts. The script installs Python packages via pip and may create/activate a venv. Installation performs network downloads and executes commands locally (expected for model-based tools).
Credentials
The skill requests no required environment variables or credentials. Optional env vars (GENOS_MODEL_PATH, GENOS_STATUS_FILE) and a config.json are documented and appropriate for configuring local model location/state. No unrelated secrets or service tokens are required by default.
Persistence & Privilege
always is false; the skill writes a local state file (./scripts/.model_loaded by default) to mark model loaded. It does not request system-wide or other-skill configuration changes. The agent can invoke it autonomously (platform default), which is expected for skills.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install genos-dna
  3. After installation, invoke the skill by name or use /genos-dna
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.1
- Changed default model status file path from ./state/.model_loaded to ./scripts/.model_loaded in documentation. - Updated references, examples, and configuration instructions to reflect the new default status file location. - Adjusted startup command: now writes status file to ./.model_loaded (project root) instead of ./state/.model_loaded. - Example config.json now points to ./scripts/.model_loaded for the state file.
v0.1.0
Genos基因序列分析技能 0.1.0 初始上线 - 新增支持使用 Genos-1.2B 基因组基础模型进行 DNA 序列分析 - 支持碱基组成统计、序列长度、预测下一个碱基与特征提取 - 提供多种输入格式,包括纯 DNA 序列、FASTA 格式等 - 新增模型状态检查与自动启动机制,支持环境变量与配置文件自定义路径 - 主要面向基因组学研究相关任务
Metadata
Slug genos-dna
Version 0.1.1
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 2
Frequently Asked Questions

What is Genos基因序列分析?

使用 Genos 模型进行 DNA 序列分析。当用户提到 DNA、基因、基因组、碱基序列、ACGT 等生物信息学相关问题时使用此技能。 It is an AI Agent Skill for Claude Code / OpenClaw, with 114 downloads so far.

How do I install Genos基因序列分析?

Run "/install genos-dna" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Genos基因序列分析 free?

Yes, Genos基因序列分析 is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Genos基因序列分析 support?

Genos基因序列分析 is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Genos基因序列分析?

It is built and maintained by flowertusi (@flowertusi); the current version is v0.1.1.

💬 Comments