Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

Ollama Updater

Ollama Updater installs or updates Ollama with curl-based breakpoint resume, auto-retry, progress display, old version cleanup, and GPU driver detection.

MIT-0 · Free to use, modify, and redistribute. No attribution required.
2 · 552 · 2 current installs · 2 all-time installs
MIT-0
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The stated purpose (install/update Ollama with resumable curl downloads) aligns with the included shell script (ollama-install.sh) and uses expected endpoints (ollama.com, GitHub raw URLs). However package metadata and docs claim additional files (main.py, a CLI wrapper 'ollama-updater', package.json 'bin' target) that are not present in the provided file manifest — this is a packaging/documentation mismatch.
!
Instruction Scope
SKILL.md and other docs instruct the agent/user to run commands that require root (sudo), create systemd units, add users/groups, and modify /usr/local/bin — all expected for an installer but high‑impact. More importantly, SKILL.md shows incorrect/ambiguous run instructions (e.g. 'bash /path/to/ollama-updater/main.py' — running bash on a .py and referencing main.py which is not present in the manifest), which is inconsistent and could lead an agent to attempt to run non-existent files or the wrong interpreter.
Install Mechanism
This is an instruction-only skill (no automated install spec). The script itself downloads from known domains (https://ollama.com and GitHub raw URLs) rather than obscure hosts. The script will write files and create systemd units when executed (expected for an installer), but there is no opaque third‑party binary download host or URL shortener in the provided content.
Credentials
No secrets or unrelated environment variables are requested. The script supports optional environment variables (OLLAMA_VERSION, OLLAMA_NO_START) but these are benign and documented. The SKILL.md does not declare these in a formal requires.env list, so the agent might not surface them clearly to users.
Persistence & Privilege
The installer intentionally requests elevated privileges (sudo/root) to install binaries, create a system user, and configure a systemd service — appropriate for a system service installer but high privilege. The skill is not always-enabled and does not request permanent automatic inclusion, which limits autonomous reach, but running it will modify system-wide settings and services.
What to consider before installing
This package appears to be an Ollama installer and its shell script behavior (resumable curl, extraction, systemd, user/group changes) is coherent with that purpose — but there are multiple red flags you should resolve before installing: 1) The manifest/docs reference files (main.py, CLI wrapper) that are not present; SKILL.md even shows 'bash .../main.py' which is wrong. This indicates sloppy packaging and increases risk of user/agent confusion. 2) The installer runs as root and will create users, write to /usr/local and /etc/systemd — review the script content (ollama-install.sh) line-by-line to confirm it only downloads from trusted hosts (ollama.com, GitHub) and makes expected changes. 3) Prefer to verify download integrity (checksums or signatures) for the Ollama binaries it fetches; consider running the script in a container or VM first if you’re unsure. 4) If you intend to let an agent invoke this skill autonomously, restrict that until you confirm the missing files and documentation inconsistencies are resolved. If you want, provide the missing files (main.py, wrapper) or a trustworthy upstream repository URL and I can re-evaluate with higher confidence.

Like a lobster shell, security has layers — review code before you run it.

Current versionv1.0.1
Download zip
latestvk976qf6ygp23zxw93p4rn2f4ch81g7qv

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

SKILL.md

ollama-updater

带断点续传功能的 Ollama 安装/更新工具,解决网络不稳定导致的下载中断问题。

功能

  • 断点续传: 使用 curl -C - 实现下载中断后继续下载
  • 自动重试: 下载失败自动重试 3 次
  • 进度显示: 实时显示下载进度
  • 智能清理: 自动清理旧版本
  • GPU 支持: 自动检测并安装 NVIDIA/AMD GPU 驱动

安装

# 使用 OpenClaw
openclaw skills install ollama-updater

# 或使用 ClawHub
clawhub install ollama-updater

使用方法

基本用法

ollama-updater

指定版本

OLLAMA_VERSION=0.5.7 ollama-updater

手动运行

# 直接运行脚本
bash /path/to/ollama-updater/main.py

# 或使用 sudo
sudo bash /path/to/ollama-updater/main.py

断点续传原理

使用 curl 的 -C - 参数实现断点续传:

# 普通下载(中断后需要从头开始)
curl -O https://example.com/file.tar.gz

# 断点续传(中断后可以继续)
curl -C - -O https://example.com/file.tar.gz

如果下载中断,只需重新运行命令,curl 会自动从断点处继续下载。

故障排查

下载总是中断

原因: 网络不稳定或服务器限速

解决:

  1. 多次运行脚本,会自动从断点续传
  2. 使用国内镜像(如有)
  3. 检查网络连接

提示 zstd 错误

原因: 需要 zstd 解压工具

解决:

# Debian/Ubuntu
sudo apt-get install zstd

# RHEL/CentOS/Fedora
sudo dnf install zstd

# Arch
sudo pacman -S zstd

权限错误

原因: 需要 sudo 权限

解决:

sudo ollama-updater

与官方脚本的区别

功能官方脚本ollama-updater
断点续传
自动重试✅ (3 次)
进度显示
GPU 检测
systemd 配置

技术细节

断点续传实现

download_file() {
    local url="$1"
    local output="$2"
    local max_retries=3
    local retry_count=0
    
    while [ $retry_count -lt $max_retries ]; do
        retry_count=$((retry_count + 1))
        
        # 使用 -C - 实现断点续传
        if curl --fail --show-error --location --progress-bar -C - -o "$output" "$url"; then
            return 0
        fi
        
        if [ $retry_count -lt $max_retries ]; then
            sleep 5
        fi
    done
    
    error "Download failed after $max_retries attempts"
}

下载流程

  1. 检查是否需要断点续传(已有部分文件)
  2. 使用 curl -C - 继续下载
  3. 下载失败自动重试(最多 3 次)
  4. 下载成功后验证并解压

系统要求

  • Linux (x86_64, aarch64)
  • macOS (x86_64, aarch64)
  • curl (必需)
  • zstd (新版 Ollama 需要)
  • sudo (系统级安装需要)

文件结构

ollama-updater/
├── main.py              # 主程序(包装官方脚本)
├── ollama-install.sh    # 改进的安装脚本
├── SKILL.md             # 技能说明
├── README.md            # 使用指南
└── package.json         # 包信息

版本历史

v1.0.0 (2026-02-20)

  • 初始版本
  • 添加断点续传功能
  • 添加自动重试机制
  • 保留官方脚本所有功能

许可证

MIT License

支持

如有问题,请提交到:https://github.com/openclaw/skills/issues

Files

7 total
Select a file
Select a file to preview.

Comments

Loading comments…