← 返回 Skills 市场
103
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install openclaw-model-manager
功能描述
OpenClaw 模型配置管理技能。用于添加、删除、更新、查看、切换、检测模型配置。当用户需要：(1) 添加新模型到配置 (2) 删除模型 (3) 更新模型参数（contextWindow、maxTokens等）(4) 查看当前模型列表 (5) 切换主模型 (6) 检查模型可用状态（测试连接）(7) 修复模型配置...
安全使用建议
This skill appears to do what it claims (edit openclaw.json, test provider connectivity), but it will read provider API keys from your OpenClaw config and make HTTP requests to provider endpoints. Before installing or invoking: 1) Back up your openclaw.json. 2) Inspect the config for provider entries and untrusted endpoints (e.g., the reference includes an unfamiliar domain wbz-api.939593.xyz). 3) Ensure you are comfortable the skill may transmit stored API keys to the listed provider URLs. 4) Prefer running connectivity tests in a controlled environment or with revoked/test keys first. 5) Confirm you have permission to restart the gateway and that restarting won't disrupt critical workflows. If you want stronger assurances, request the author/source or a signed/official package and ask them to declare required config paths in the metadata.
功能分析
Type: OpenClaw Skill
Name: openclaw-model-manager
Version: 1.0.1
The skill is designed to manage and test OpenClaw model configurations, which requires reading and modifying 'openclaw.json' (containing sensitive API keys) and executing PowerShell commands via 'exec' to perform network connectivity tests. While these high-risk capabilities are aligned with the stated purpose of a model manager, the combination of credential access and outbound network requests to various endpoints—including a third-party proxy domain 'wbz-api.939593.xyz'—presents a significant attack surface. No clear evidence of intentional malice or data exfiltration was found in SKILL.md or the reference files.
能力评估
Purpose & Capability
The name/description align with the instructions: adding/removing/updating models, switching primary model, and testing connectivity. However, the SKILL.md explicitly references a local config path (C:\Users\Administrator\.openclaw\openclaw.json) while the registry metadata lists no required config paths — a mild inconsistency in metadata vs. runtime expectations.
Instruction Scope
Runtime instructions tell the agent to read the OpenClaw configuration (including provider API keys), modify it, restart the gateway, and perform parallel HTTP test requests to each provider. These actions are within scope for a model manager, but they include reading secrets from local config and making outbound network calls to provider endpoints — both are sensitive behaviors the user should expect and approve.
Install Mechanism
Instruction-only skill with no install spec and no code files — lowest install risk. Nothing is downloaded or written by an installer.
Credentials
The skill does not request environment variables, but it instructs reading provider API keys from the local openclaw.json. Access to those API keys is required for connectivity testing, so the behavior is proportionate to the purpose, but the metadata omits declaring required config paths and secrets. Additionally, references/model-info.md lists provider endpoints; at least one domain (https://wbz-api.939593.xyz) is an unrecognized third-party domain and should be vetted before any tests that would send stored API keys to it.
Persistence & Privilege
No elevated persistence requested (always: false). The skill instructs use of gateway(action=restart) to apply changes, which is normal for configuration changes but requires appropriate privileges; nothing indicates the skill will modify other skills or system-wide settings beyond the OpenClaw config.
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 openclaw-model-manager - 安装完成后，直接呼叫该 Skill 的名称或使用
/openclaw-model-manager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
v1.0.1
v1.0.1 增强说明：
- 补全和丰富了使用说明，新增适用场景、核心能力、常见用户请求示例等说明。
- 明确了模型状态检测的并行测试方式、排序规则和分组逻辑。
- 优化错误码处理说明，补充`401/403/429/400/5xx`等常见原因解释。
- 增加了主模型切换前的检查建议，强调修改后需重启生效。
- 进一步规范了文档结构和提示信息，便于日常维护和故障排查。
v1.0.0
OpenClaw Model Manager v1.0.0 — Initial Release
- 提供模型配置的添加、删除、更新、查看、切换和检测功能
- 支持并行检测多个模型可用性并按能力排序输出
- 更新模型参数（如 contextWindow, maxTokens）支持
- 切换默认主模型和修复配置问题
- 明确文档工作流程与常用配置说明
元数据
常见问题
Model Manager 是什么？
OpenClaw 模型配置管理技能。用于添加、删除、更新、查看、切换、检测模型配置。当用户需要：(1) 添加新模型到配置 (2) 删除模型 (3) 更新模型参数（contextWindow、maxTokens等）(4) 查看当前模型列表 (5) 切换主模型 (6) 检查模型可用状态（测试连接）(7) 修复模型配置... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 103 次。
如何安装 Model Manager？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install openclaw-model-manager」即可一键安装，无需额外配置。
Model Manager 是免费的吗？
是的，Model Manager 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Model Manager 支持哪些平台？
Model Manager 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Model Manager？
由 the2015（@the2015）开发并维护，当前版本 v1.0.1。
推荐 Skills