PansModel Manager

PanSclaw 模型管理器 - 管理自定义模型接入和切换。 支持添加、删除、切换、列出模型配置。 触发关键词: 模型管理, 切换模型, 添加模型, 删除模型, 模型列表, 配置模型, 更换模型, 接入模型, 模型设置

Security Scan

The skill does what it says — it reads/writes a local OpenClaw config to add, list, switch, and delete model provider entries; nothing in the bundle requests unrelated credentials or network execution.

Purpose & Capability

Name/description match the actual behavior. The script and SKILL.md both implement adding/removing/switching/listing model providers and map provider names to API adapters; required capabilities are proportional to model-management.

Instruction Scope

Runtime instructions and the script operate on a single, well-defined config path (~/.openclaw-pansclaw/openclaw.json). They do not reference other system files, environment variables, or remote endpoints for execution. The SKILL.md explicitly documents the commands and the config location.

Install Mechanism

No install spec is provided (instruction-only skill with an included Python helper). Nothing in the manifest downloads remote code or creates system-wide binaries; the included Python script is small and readable.

Credentials

The skill does not request environment variables or external credentials, but it explicitly accepts and writes API keys provided by the user into the config file (openclaw.json) in the user's home directory. Storing API keys in plaintext in a local config is expected for this tool but is a sensitivity consideration for users.

Persistence & Privilege

The skill modifies a user-local configuration file under ~/.openclaw-pansclaw/openclaw.json (its own scope). always is false and it does not modify other skills or global agent settings beyond that config file. This behavior is appropriate for a model manager but does mean the skill will persistently change agent behavior via that file.

Assessment

This skill appears internally consistent for managing OpenClaw model provider entries. Before installing or using it: (1) Review the included script (scripts/PansModel-Manager.py) yourself — it is small and readable. (2) Be aware that when you 'add' a provider you supply API keys which the script will save in ~/.openclaw-pansclaw/openclaw.json in plaintext — consider file permissions (chmod 600) or using short-lived keys. (3) Only add API keys for providers you trust; if unsure, create/revoke keys after testing. (4) Back up your existing openclaw.json before using the tool so you can restore previous settings. If you want stronger guarantees, request that the author implement encrypted secret storage or use environment-based secrets instead.

模型管理器

功能

管理 PanSclaw 的模型配置，支持：

• 添加自定义模型提供商
• 切换默认模型
• 列出已配置模型
• 删除模型配置

使用方法

添加模型

/模型 添加 厂商=<name> 地址=<url> 密钥=<key> 模型=<model_id>

示例：

/模型 添加 厂商=deepseek 地址=https://api.deepseek.com/v1 密钥=sk-xxx 模型=deepseek-chat

切换模型

/模型 切换 <厂商>/<模型名>

示例：

/模型 切换 deepseek/deepseek-chat

列出模型

/模型 列表

删除模型

/模型 删除 <厂商>

支持的 API 类型

厂商	API 适配器
openai	openai-completions
anthropic	anthropic-messages
deepseek	openai-completions
moonshot	openai-completions
zhipu	openai-completions
qwen	openai-completions
ollama	ollama

配置文件位置

~/.openclaw-pansclaw/openclaw.json

配置修改后需要重启 Gateway 生效。