model-platform-guide
v1.0.0查询主流大模型开放平台的 API 文档入口与精简调用要点,适用于月之暗面 Kimi、阿里云百炼、硅基流动、DeepSeek、OpenRouter 等平台。用于回答鉴权方式、base URL、OpenAI 兼容调用、聊天补全、模型名、常见差异、多平台对比与文档入口。当用户提到 Kimi、Moonshot、百炼、Da...
Security Scan
Capability signals
These labels describe what authority the skill may exercise. They are separate from suspicious or malicious moderation verdicts.
OpenClaw
Benign
high confidencePurpose & Capability
Skill name/description match the shipped artifacts: it contains concise reference files, migration recipes, curl/Python/Node examples and an index of official doc URLs for the listed platforms. There are no unrelated binaries, installs, or environment variables required by the skill itself.
Instruction Scope
The SKILL.md explicitly limits itself to: reading its own references/* files, returning links and short examples, and performing 'dynamic verification' by visiting official docs when needed. Those actions are appropriate for a documentation/navigation skill. The instructions do not tell the agent to read arbitrary system files, secrets, or unrelated config paths. Note: dynamic verification implies the agent may fetch external documentation pages at runtime — this is expected behavior for live checks.
Install Mechanism
No install spec and no code files that would be written/executed. Instruction-only skills are lowest risk from an install perspective.
Credentials
The reference files include example environment variable names (e.g., MOONSHOT_API_KEY, DASHSCOPE_API_KEY) solely as usage examples for calling remote APIs. The skill does not declare any required env vars or request credentials. There is no disproportionate credential access.
Persistence & Privilege
always:false and no special privileges are requested. The skill does not attempt to persist configuration outside its own references or alter other skills. Autonomous invocation is allowed (platform default) but not combined with other red flags here.
Assessment
This skill is a local collection of concise references and call examples and appears coherent with its described purpose. Things to consider before installing: (1) The agent may perform live web checks of the official docs when you ask about up-to-date model capabilities — if you care about privacy/networking, be aware of that behavior. (2) The included files show example environment variable names and code snippets; the skill itself does not require or collect keys, but you should never paste actual secret keys into chat or into code examples when using the skill. (3) Validate any official URLs returned by the skill before using them, and prefer copying examples into a secure environment (your own code) rather than executing unreviewed snippets. If you want greater assurance, ask the publisher for provenance of the references or review a few of the referenced URLs manually to confirm they point to the expected official documentation.Like a lobster shell, security has layers — review code before you run it.
latest
model-platform-guide
这个 skill 用来处理“模型开放平台 API 文档”相关工作。它采用链接优先、摘要精简、按需访问官方文档的方式,避免 skill 自身过于臃肿,也减少文档过期风险。
当前覆盖平台
- 月之暗面 / Moonshot / Kimi
- 阿里云百炼 / DashScope / Model Studio
- 硅基流动 / SiliconFlow
- DeepSeek
- OpenRouter
- 智谱开放平台 / BigModel
- 火山方舟 / Volcengine Ark
- 腾讯混元
- 百度千帆
- MiniMax
主入口
默认先看:references/index.md
它现在是整个 skill 的能力入口总表,优先用于定位:
- 平台首页
- 模型列表 / 模型概览
- Chat / Responses / Tools / MCP
- Embeddings / Rerank
- Release Notes / 更新入口
适用场景
当用户需要以下任一内容时使用:
- 查询某个平台的 API 调用方式
- 对比多个平台的 base URL、鉴权、模型名、SDK 兼容性
- 生成 Python / Node.js / curl 的调用示例
- 解释 OpenAI 兼容接口怎么改造
- 汇总常见错误、流式输出、工具调用、多模态能力
- 编写“模型平台 API 文档”或“接入指南”
工作方式
- 先判断用户要的是:
- 单个平台文档
- 多平台对比
- 接入示例
- 现有代码迁移
- 最新模型/能力变化核验
- 默认先查看
references/index.md,定位到对应平台和能力入口。 - 优先给出:
- 官方文档入口
- base URL
- 认证方式
- 常用 endpoint
- 最小可运行示例
- 只在必要时读取对应
references/*.md。 - 如果 skill 中信息不够或可能过期,优先现场访问官方文档链接,而不是依赖旧摘要。
- 如果问题涉及“现在是否支持”“最近新增了什么”“某模型是否具备某能力”,默认按
references/live-check-policy.md执行动态核验。 - 如果问题是“模型 / 联网搜索 / embeddings”这类高变能力问题,优先走
references/model-search-embedding-checklist.md。 - 回复中可以明确说明:skill 内保存的是精简导航和常用要点,官方文档才是最终依据。
参考文件
平台摘要
- 月之暗面:
references/moonshot.md - 阿里云百炼:
references/aliyun-bailian.md - 硅基流动:
references/siliconflow.md - DeepSeek:
references/deepseek.md - OpenRouter:
references/openrouter.md - 智谱:
references/bigmodel.md - 火山方舟:
references/ark.md - 腾讯混元:
references/hunyuan.md - 百度千帆:
references/qianfan.md - MiniMax:
references/minimax.md
横向比较
- 多平台对比:
references/comparison.md - 能力矩阵:
references/capability-matrix.md
动态核验
- 文档入口总表:
references/index.md - 动态核验策略:
references/live-check-policy.md - 模型/联网搜索/Embeddings 核验清单:
references/model-search-embedding-checklist.md
调用模板
- OpenAI 迁移模板:
references/openai-migration.md - Python 模板:
references/python-recipes.md - Node.js 模板:
references/nodejs-recipes.md - curl 模板:
references/curl-recipes.md
输出偏好
默认输出:
- 先给结论
- 再给官方文档链接
- 再给最小可运行示例
- 最后给差异对比或注意事项
如果用户要“写文档/整理文档”,优先输出结构化 Markdown:
- 平台简介
- 官方文档入口
- 鉴权
- Base URL
- 核心接口
- 请求示例
- 响应示例
- 常见坑
- 对比表
如果用户明确要代码模板:
- Python →
references/python-recipes.md - Node.js →
references/nodejs-recipes.md - curl →
references/curl-recipes.md - OpenAI 改造/迁移 →
references/openai-migration.md
如果用户明确要平台选型、能力横向比较:
- 先读
references/capability-matrix.md - 再按需看
references/comparison.md - 遇到高变能力再结合
references/live-check-policy.md
如果用户明确问的是“模型 / 联网搜索 / embeddings”:
- 先读
references/model-search-embedding-checklist.md - 再结合
references/index.md做现场核验
维护规则
- 不要把整站原文硬塞进 SKILL.md。
references/只保留高频字段、最小示例、官方链接和注意事项。- 当新增平台时,优先新增一个精简 reference 文件,而不是镜像全站文档。
- 如果文档 URL 失效,优先更新
references/index.md和对应平台入口。 - 对于模型能力、最新模型、价格、preview/deprecated 状态等高变信息,不要把本地摘要当成最终依据;优先按
references/live-check-policy.md动态核验。
Comments
Loading comments...
