← 返回 Skills 市场
761
总下载
0
收藏
11
当前安装
1
版本数
在 OpenClaw 中安装
/install ollama
功能描述
Run, tune, and troubleshoot local Ollama models with reliable API patterns, Modelfiles, embeddings, and hardware-aware deployment workflows.
安全使用建议
This skill appears coherent for managing local Ollama models. Before installing or using it: 1) confirm you have the official 'ollama' binary on the machine (do not run untrusted installers); 2) review and explicitly approve any writes to ~/ollama/ (the skill stores durable operational notes there); 3) keep port 11434 bound to localhost unless you explicitly approve and configure auth/firewalls; 4) the skill may run local commands (ollama list/serve/run/etc.) — only grant access on machines you control. If you plan to use any cloud paths, verify the skill asks you first and inspect the exact endpoints before enabling them.
功能分析
Type: OpenClaw Skill
Name: ollama
Version: 1.0.0
The Ollama skill bundle is designed to manage local LLM runtimes, focusing on installation, configuration, and troubleshooting. It uses a local directory (~/ollama/) for state persistence and interacts primarily with the local Ollama API (127.0.0.1:11434). The instructions in SKILL.md and setup.md emphasize user consent for remote exposure and data persistence, and all external network activity is directed to official endpoints (ollama.com) for documentation and model pulls. No evidence of data exfiltration, malicious execution, or unauthorized persistence was found.
能力评估
Purpose & Capability
Name/description (local Ollama model management) align with required binaries (ollama) and optional tools (curl, jq). Required config paths (~/ollama/, ~/.ollama/) are consistent with storing durable local state for this purpose.
Instruction Scope
Runtime instructions focus on local CLI and local HTTP API calls (127.0.0.1:11434), Modelfile workflows, embeddings, and persistent notes under ~/ollama/. The skill explicitly instructs asking for user consent before creating memory files; nevertheless, any install operator should confirm before allowing writes to the home-directory memory.
Install Mechanism
Instruction-only skill with no install spec and no remote download/extract steps. Lowest-risk install posture (relies on existing 'ollama' binary and standard CLI tools).
Credentials
No environment variables or credentials requested. The only nontrivial access is to user config paths in the home directory (~/ollama/, ~/.ollama/) for durable state — this is proportional to the stated purpose but is a persistence/privacy consideration the user should accept explicitly.
Persistence & Privilege
always:false (not force-included). The skill can be invoked autonomously (platform default) but does not request elevated persistence or modify other skills. It documents guardrails around remote exposure and requests explicit approval for non-local binding or service-manager changes.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install ollama - 安装完成后，直接呼叫该 Skill 的名称或使用
/ollama触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
Initial release with local model setup, stable JSON output, custom model workflows, safer remote access, and recovery playbooks.
元数据
常见问题
Ollama 是什么？
Run, tune, and troubleshoot local Ollama models with reliable API patterns, Modelfiles, embeddings, and hardware-aware deployment workflows. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 761 次。
如何安装 Ollama？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install ollama」即可一键安装，无需额外配置。
Ollama 是免费的吗？
是的，Ollama 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Ollama 支持哪些平台？
Ollama 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（linux, darwin, win32）。
谁开发了 Ollama？
由 Iván（@ivangdavila）开发并维护，当前版本 v1.0.0。
推荐 Skills