中文 EN
Start Review
Skill Trust Decision

image-generation-guide

本地AI图像生成指南技能，脚本会无警告删除用户主目录PNG文件，curl|sh安装方式存在风险，已在skill-card.md声明但SKILL.md未提及。

Install decision first Source: ClawHub Scanned: Jun 24, 2026
Files 6
Artifacts 7
Violations 1
Findings 4
Most direct threat evidence
High Sensitive Access
脚本删除用户主目录PNG文件

scripts/flux-gen 第35行执行 rm -f "$HOME"/*.png 会无警告删除用户主目录下的所有PNG图像，可能导致用户重要数据丢失

scripts/flux-gen:35

Why this conclusion was reached

3/4 dimensions flagged
Block
Declared vs actual capability

1 undeclared or violating capabilities were inferred.

Block
Hidden execution and egress

1 high-risk artifacts or egress signals were extracted.

Block
Attack chain and severe findings

The report includes 0 attack-chain steps and 2 severe findings.

Review
Dependencies and supply chain hygiene

Dependency information is incomplete, so supply-chain confidence stays limited.

What drove the risk score up

脚本无警告删除用户主目录PNG +20

scripts/flux-gen:35 执行 rm -f "$HOME"/*.png 会删除用户重要文件

curl|sh远程脚本执行 +15

guide/full-guide.md:71 包含 curl -fsSL https://ollama.com/install.sh | sh

文档声明不完整 +10

SKILL.md未声明文件删除行为，仅在skill-card.md有说明

Most important evidence

High Sensitive Access

脚本删除用户主目录PNG文件

scripts/flux-gen 第35行执行 rm -f "$HOME"/*.png 会无警告删除用户主目录下的所有PNG图像，可能导致用户重要数据丢失

scripts/flux-gen:35
删除此命令，仅保留 $WORK_DIR/*.png 的清理，或在脚本开头添加用户确认提示
High RCE

curl-to-shell安装命令

guide/full-guide.md 包含 curl -fsSL https://ollama.com/install.sh | sh 远程脚本执行，存在供应链风险

guide/full-guide.md:71
建议使用官方包管理器（brew/apt）或手动下载验证后执行
Medium Doc Mismatch

SKILL.md与skill-card.md风险声明不一致

SKILL.md未声明文件删除风险，仅在skill-card.md的Known Risks部分说明，造成文档信息不对称

SKILL.md:1
SKILL.md应添加风险警告部分，与skill-card.md保持一致
Low Supply Chain

使用非官方Ollama模型源

脚本引用 x/flux2-klein:4b 等非官方模型路径，可能来自第三方，存在模型供应链风险

scripts/flux-gen:5
确认模型来源可信，或使用ollama.com官方模型

Declared capability vs actual capability

Shell Pass
Declared WRITE
→
Inferred WRITE
scripts/flux-gen:38 调用 ollama run
Filesystem Block
Declared WRITE
→
Inferred WRITE
scripts/flux-gen:35 意外删除 $HOME/*.png 但SKILL.md未声明
Network Pass
Declared READ
→
Inferred NONE
未发现网络行为，仅文档链接

Suspicious artifacts and egress

Critical Dangerous Command
curl -fsSL https://ollama.com/install.sh | sh

guide/full-guide.md:71

Medium External URL
https://ollama.com/install.sh

guide/full-guide.md:71

Medium External URL
https://discord.gg/ollama

guide/full-guide.md:301

Medium External URL
https://ollama.com/blog/image-generation

guide/full-guide.md:342

Medium External URL
https://blackforestlabs.ai/

guide/full-guide.md:343

Medium External URL
https://clawhub.ai/user/amrree

skill-card.md:7

Medium External URL
https://clawhub.ai/amrree/skills/image-gen-guide

skill-card.md:27

Dependencies and supply chain

There are no structured dependency warnings.

File composition

6 files · 601 lines
Markdown 5 files · 549 linesOther 1 files · 52 lines
Files of concern · 4
guide/full-guide.md Markdown · 348 lines
curl-to-shell安装命令 · curl -fsSL https://ollama.com/install.sh | sh · https://ollama.com/install.sh · https://discord.gg/ollama · https://ollama.com/blog/image-generation · https://blackforestlabs.ai/
skill-card.md Markdown · 45 lines
https://clawhub.ai/user/amrree · https://clawhub.ai/amrree/skills/image-gen-guide
scripts/flux-gen Other · 52 lines
脚本删除用户主目录PNG文件 · 使用非官方Ollama模型源
SKILL.md Markdown · 40 lines
SKILL.md与skill-card.md风险声明不一致
Other files · prompts.md · README.md

Security positives

skill-card.md明确声明了已知风险（PNG删除和curl|sh），提供了缓解建议
脚本功能与文档描述基本一致，无隐藏的凭证收割或数据外泄行为
权限声明与实际需要基本匹配（shell.exec用于调用ollama命令）
无base64编码、eval执行、敏感路径遍历等恶意行为特征