Agent Platform Rules
Cited, dated answers: may an AI agent post, sign up or automate on Reddit, HN, Bluesky, GitHub, X?
我该使用它吗
质量与安全性
A
基于对工具定义和协议合规性的自动分析。
上下文开销
~233token 数（工具定义）
~461 B典型响应大小
对注意力的影响极小（占 128k 上下文窗口的 0.18%）
这是每次将服务器的工具加载到模型上下文窗口时所消耗的大致 token 数。数值越高，可用于其他任务的注意力就越少。
安装
一键安装
将以下内容添加到你的 `claude_desktop_config.json` 文件中：
{
"mcpServers": {
"agent-platform-rules": {
"url": "https://4fucgqu5if.execute-api.us-east-1.amazonaws.com/mcp"
}
}
}远程端点
https://4fucgqu5if.execute-api.us-east-1.amazonaws.com/mcpstreamable-http它能做什么
工具清单
工具（2）
🟢 只读🟡 写入🔴 删除⚪ 未知
🟢list_platforms
List the platforms this server has verified rules for, with each one's last-verified date and a one-line answer to "is automation allowed?".
输入模式
{
"type": "object",
"properties": {},
"additionalProperties": false
}🟢get_platform_rules(platform)
Before signing up, posting or commenting on a platform as an AI agent or automated account, call this to see whether automation and AI-generated text are allowed, what must be disclosed (e.g. labels), verbatim quotes from the rules, source URLs, and when they were last verified. Not legal advice; open the sources before acting.
输入模式
{
"type": "object",
"properties": {
"platform": {
"type": "string",
"description": "One of: hackernews, reddit, bluesky, mastodon.social, github, x (aliases like \"hn\", \"twitter\", \"bsky\" work)."
}
},
"required": [
"platform"
],
"additionalProperties": false
}社区
证据
最近观测
已验证未记录版本2 个工具