skill
ipure-skills
Agent Skills for checking IP purity & proxy quality — score, residential vs datacenter, proxy/VPN/Tor, blocklists, suitability for ChatGPT, X sign-up, streaming, gaming, e-commerce. Free IPure API, no
Set up
Skills are installed by placing their folder where your agent loads skills.
The download comes from github.com/minner-fun/ipure-skills itself, not from our server, so what you get is exactly what its author published.
Install instructions come from the README
README
synced Oct 22, 2026IPure Skills — IP purity check for AI agents
Agent Skills that teach Claude Code, Codex CLI and other skill-aware agents to answer one question well:
Is this IP clean enough for what I'm about to do?
They call the free IPure API — no key, no sign-up — and turn the result into an answer a person can act on: a 0-100 purity score, residential vs datacenter, native vs broadcast, proxy / VPN / Tor detection (with the VPN operator when known), blocklists, abuse and attack history, and per-scenario suitability for AI services, social sign-up, streaming, gaming platforms, cross-border e-commerce and email. Every deducted point is traceable to a signal and a data source, and the skill always states what an IP check cannot tell you.
Skills
| Skill | Language | Use it for |
|---|---|---|
ip-purity-check | English | "Is this proxy clean?", "Can I use this IP for ChatGPT / X sign-up / Netflix / Steam?", "Which of these IPs is best?", pre-flight check of an agent's egress IP |
ip-purity-check-zh | 中文 | 「这个 IP 干不干净」「这个节点能不能登 ChatGPT」「是不是原生 IP / 住宅 IP」「这几个 IP 用哪个」 |
MCP server
The same checks are available as a hosted MCP server at https://ipure.dev/mcp (Streamable HTTP, no key). It exposes three tools:
inspect_ip— score, type, flags, blocklists, top signals, scenarios and unknowns for one IPassess_ip_for_use_case— whether one IP fits AI services, social sign-up, streaming, gaming, e-commerce or emailcompare_egress_ips— rank 2-5 candidate IPs, optionally for one use case
The plugin includes .mcp.json, so installing it also registers the server. For a manual setup:
claude mcp add --transport http ipure https://ipure.dev/mcp
MCP tools return a compact summary plus report_url; use the HTTP JSON API when you need every field.
Install
Claude Code — as a plugin
/plugin marketplace add minner-fun/ipure-skills
/plugin install ipure@ipure-skills
Claude Code — copy the skill
git clone https://github.com/minner-fun/ipure-skills
cp -r ipure-skills/skills/ip-purity-check ~/.claude/skills/ # personal
# or into a project: cp -r ipure-skills/skills/ip-purity-check .claude/skills/
Codex CLI
git clone https://github.com/minner-fun/ipure-skills
cp -r ipure-skills/skills/ip-purity-check ~/.codex/skills/
Any agent that reads SKILL.md files works the same way: drop the folder where your agent looks for skills. The only requirement is curl (and ideally jq).
Try it
> I bought a proxy, exit IP is 146.70.132.85. Can I use it for ChatGPT?
> Check my current IP before we start scraping.
> Which of these is best for registering X accounts: 1.2.3.4, 5.6.7.8, 9.10.11.12
Or call the API yourself:
curl -s "https://ipure.dev/en/ip/8.8.8.8.json" | jq '{purity: .risk.purity, verdict: .risk.verdict, scenarios: [.scenarios[] | {id, score, levelLabel}], report: .reportUrl}'
No agent? Use it from a terminal
The same check is available as plain text, so you can run it on a fresh VPS before anything is installed:
curl -L ipure.dev # the machine's current egress IP
curl -L ipure.dev/cli/8.8.8.8 # a specific IP (IPv4 or IPv6)
bash <(curl -sL ipure.dev/cli.sh) # IPv4 and IPv6, one report each
-L follows the redirect to https. Add ?color=0 (or NO_COLOR=1 for the script) to drop ANSI colours before pasting the report somewhere. Nothing is installed and nothing runs on your machine beyond curl; cli.sh is about twenty lines and worth reading first. The text report is written in Chinese — agents should keep using the JSON API above.
What the API gives you
risk.purity(0-100, higher is cleaner), level, verdict, and a separateconfidencerisk.factors[]— where every point came fromusageType,nativeType,flags(proxy / VPN / Tor / hosting / relay),vpnOperatorscenarios[]—ai,social,streaming,gaming,ecommerce,email, each weighted differentlyblocklists[],abuse[](reports and attack history),exposure,sharingfeedback— real-world experience ratings (1-5) per scenario from people who were actually using the IPsources[]— which data sources answered and which failed (a failed source is not a clean result)scenarioApplicable/scenarioNote—falsewhen the address is public infrastructure (e.g. 8.8.8.8) rather than an egress IP, so scenario suitability does not applyunknowns[]— what cannot be determined from an IP alonereportUrl— the human-readable report to cite
Docs: https://ipure.dev/en/docs/api · OpenAPI: https://ipure.dev/openapi.json · For models: https://ipure.dev/en/llms.txt · Example report: https://ipure.dev/en/ip/8.8.8.8 · Aggregate statistics (Chinese): https://ipure.dev/insights
Every report exists in three shapes and two languages: /en/ip/{ip} (web page), /en/ip/{ip}.json, /en/ip/{ip}.txt — drop /en for Chinese. /api/lookup?ip={ip}&lang=en is equivalent to the .json address.
Data & guides
- IP purity data report, Q4 2026 — 56,688 real checks, with method, sample bias and limitations (CC BY 4.0).
- Why streaming IPs get "wrong region" and what an IP check can say about social sign-up.
- Background concepts: ASN / CIDR and IP sharing.
Limits
Free, no API key. 30 requests per minute per client. IPs already in IPure's database return instantly; an IP that is not yet in the database triggers a live lookup, and unverified clients get 5 of those per source IP per day. When that runs out the API returns 403 verification_required with a reportUrl — open it in a browser once and the IP is in the database from then on. Looking up your own egress IP (/api/lookup with no ip) never needs verification. Please do not bulk-scan ranges.
Honest boundaries
An IP check describes IP-level risk from public and partner data at query time. It cannot see a platform's internal reputation data, whether an IP is truly exclusive, which accounts used it before, or your device fingerprint and behaviour. The skills are written to say so every time, and never to promise that an IP is "safe" or "won't be flagged".
中文说明
这是给 AI Agent 用的「IP 纯净度检测」技能,支持 Claude Code、Codex CLI 等能读取 SKILL.md 的 Agent。它调用免费的 IPure 接口(无需 key、无需注册),回答一个问题:这个 IP 拿去做我要做的事,够不够干净?
返回 0-100 纯净度评分、住宅 / 机房识别、原生 / 广播判定、代理 / VPN / Tor 判定(已知商业 VPN 给出运营商)、黑名单、滥用与攻击记录,以及 AI 服务、社交平台注册、流媒体、游戏平台、跨境电商、邮件发送六个场景各自的适用性。每一分的扣分依据和数据来源都可追溯,并且每次都会说明「仅凭 IP 无法判断的事」。
也可以走托管的 MCP 服务器 https://ipure.dev/mcp(Streamable HTTP),工具为 inspect_ip、assess_ip_for_use_case、compare_egress_ips。插件自带 .mcp.json,安装插件即注册;单独加:
claude mcp add --transport http ipure https://ipure.dev/mcp
安装(Claude Code):
/plugin marketplace add minner-fun/ipure-skills
/plugin install ipure@ipure-skills
或直接复制技能目录:
git clone https://github.com/minner-fun/ipure-skills
cp -r ipure-skills/skills/ip-purity-check-zh ~/.claude/skills/
然后直接问:「我买了个代理,出口是 146.70.132.85,能登 ChatGPT 吗?」「开始爬之前先查一下我现在的出口 IP」。
不用 Agent 也能查,终端里一条命令(新机到手、什么都没装的时候最顺手):
curl -L ipure.dev # 查本机当前的出口 IP
curl -L ipure.dev/cli/8.8.8.8 # 查指定 IP
bash <(curl -sL ipure.dev/cli.sh) # IPv4、IPv6 各查一次
加 ?color=0 关掉颜色,方便贴到论坛。不安装任何东西,cli.sh 只有二十来行,建议先读一遍。
接口文档:https://ipure.dev/docs/api · 数据洞察:https://ipure.dev/insights
License
MIT