feat: clarify Cloudflare reference transfers for v0.9.0

.github/ISSUE_TEMPLATE/bug_report.yml

@@ -12,7 +12,7 @@ body:
     id: version
     attributes:
       label: ipcheck version
-      placeholder: ipcheck 0.8.2
+      placeholder: ipcheck 0.9.0
     validations:
       required: true
   - type: dropdown

CHANGELOG.md

@@ -5,6 +5,28 @@ All notable changes to ipcheck are documented here. The project follows
 
 ## [Unreleased]
 
+## [0.9.0] - 2026-07-23
+
+### Changed
+
+- Rename the capped Cloudflare section to proxy-path reference transfers and
+  state directly in terminal, Markdown, JSON, and bilingual documentation that
+  it is not a peak-bandwidth test or AI API throughput measurement.
+- Replace readiness-score rule v2 with rule v3: the measured AI service path
+  now supplies all positive points, while complete low Cloudflare samples can
+  only deduct two points per direction and incomplete samples deduct one.
+- Describe Cloudflare results as high, moderate, or low samples instead of
+  presenting them as general-purpose fast/adequate/slow network ratings.
+- Keep the existing JSON `bandwidth` object for compatibility while adding
+  explicit scope, method, API-path, and peak-bandwidth metadata.
+
+### Fixed
+
+- Rate reference samples using the same rounded Mbps value shown to users so a
+  displayed threshold value cannot contradict its label.
+- Remove duplicate Cloudflare warnings from the primary AI-service diagnosis
+  and keep the auxiliary signal in its own clearly scoped section.
+
 ## [0.8.2] - 2026-07-23
 
 ### Fixed
@@ -149,7 +171,8 @@ All notable changes to ipcheck are documented here. The project follows
 - Homebrew and direct-download packaging.
 - Median/P95 TTFB, jitter, reference bandwidth, and macOS `networkQuality`.
 
-[Unreleased]: https://github.com/jacklv-coder/ipcheck/compare/v0.8.2...HEAD
+[Unreleased]: https://github.com/jacklv-coder/ipcheck/compare/v0.9.0...HEAD
+[0.9.0]: https://github.com/jacklv-coder/ipcheck/compare/v0.8.2...v0.9.0
 [0.8.2]: https://github.com/jacklv-coder/ipcheck/compare/v0.8.1...v0.8.2
 [0.8.1]: https://github.com/jacklv-coder/ipcheck/compare/v0.8.0...v0.8.1
 [0.8.0]: https://github.com/jacklv-coder/ipcheck/compare/v0.7.0...v0.8.0

README.md

@@ -11,7 +11,7 @@ proxy before it breaks your coding flow.
 
 `ipcheck` is a zero-dependency Bash CLI that tests the real service routes used
 by AI coding clients. It reports reachability, median/P95 time to first byte,
-jitter, reference download/upload speed, and a plain-language readiness score.
+jitter, capped proxy-path transfer samples, and a plain-language readiness score.
 No Codex or Claude login is required. It never reads API keys, sends prompts,
 or makes billable model requests.
 
@@ -35,7 +35,7 @@ brew upgrade ipcheck
 
 ```bash
 mkdir -p "$HOME/.local/bin"
-curl -fsSL https://raw.githubusercontent.com/jacklv-coder/ipcheck/v0.8.2/bin/ipcheck \
+curl -fsSL https://raw.githubusercontent.com/jacklv-coder/ipcheck/v0.9.0/bin/ipcheck \
   -o "$HOME/.local/bin/ipcheck"
 chmod +x "$HOME/.local/bin/ipcheck"
 ```
@@ -50,7 +50,7 @@ Make sure `$HOME/.local/bin` is on `PATH`.
 | --- | --- |
 | Speedtest | Peak bandwidth to a nearby test server |
 | `ping` / `curl` | Basic reachability to one address |
-| `ipcheck` | Codex/Claude routes, proxy path, TTFB, P95, jitter, failures, bandwidth, and coding readiness |
+| `ipcheck` | Codex/Claude routes, proxy path, TTFB, P95, jitter, failures, capped reference transfers, and coding readiness |
 
 For coding agents, a 500 Mbps connection can still feel slow when the first byte
 takes five seconds or the proxy route is unstable. `ipcheck` measures those
@@ -62,8 +62,8 @@ interactive bottlenecks directly.
   broken one.
 - Understands OpenAI, Anthropic, LiteLLM-style gateways, and Alibaba Cloud
   Model Studio/DashScope Anthropic-compatible routes.
-- Explains whether the main problem is reachability, TTFB, P95, jitter,
-  download speed, or upload speed.
+- Explains whether the main problem is reachability, TTFB, P95, jitter, or a
+  low Cloudflare reference-transfer sample.
 - Produces human, Markdown, and stable versioned JSON reports.
 - Shows animated progress, adapts to narrow terminals, and exits cleanly with
   status `130` when cancelled with `Ctrl+C`.
@@ -87,14 +87,14 @@ or Mantle. `ipcheck` warns instead of silently testing the wrong provider.
 | Command | Purpose |
 | --- | --- |
 | `ipcheck` | Auto-detect installed clients and run the standard check |
-| `ipcheck --quick` | One sample, shorter timeout, no bandwidth transfer |
+| `ipcheck --quick` | One sample, shorter timeout, no reference transfer |
 | `ipcheck codex` / `ipcheck claude` | Test only one client |
 | `ipcheck all` | Test Codex and Claude Code together |
 | `ipcheck --explain-score` | Show every score component |
 | `ipcheck --json` | Emit structured output for automation |
 | `ipcheck --markdown` | Create a shareable support report |
 | `ipcheck --system` | Add macOS `networkQuality` measurements |
-| `ipcheck --no-bandwidth` | Skip capped Cloudflare download/upload checks |
+| `ipcheck --no-bandwidth` | Skip capped Cloudflare reference transfers |
 
 Run `ipcheck --help` for every option and environment variable.
 
@@ -115,8 +115,13 @@ TTFB is measured from a credential-free protocol request. It reflects DNS,
 proxy, TLS, network, and gateway ingress—not authenticated model generation or
 time to the model's first token.
 
-The 0–100 score is a transparent rule, not a user percentile. It prioritizes
-service reachability and latency over peak bandwidth. See
+The Cloudflare samples describe small transfers from the current proxy/network
+path to Cloudflare. They are not peak-bandwidth tests and do not represent
+OpenAI, Anthropic, GitHub, or npm throughput.
+
+The 0–100 score is a transparent rule, not a user percentile. The measured AI
+service path supplies all positive points; low Cloudflare samples can only make
+a small deduction. See
 [Scoring and result rules](docs/scoring.md) for the exact calculation.
 
 ## Claude Code gateway example
@@ -149,7 +154,7 @@ See [SECURITY.md](SECURITY.md) for private vulnerability reporting.
 ## Documentation
 
 - [Scoring and result rules](docs/scoring.md)
-- [Routes, proxies, and bandwidth](docs/network.md)
+- [Routes, proxies, and reference transfers](docs/network.md)
 - [Reports, automation, and exit codes](docs/reporting.md)
 - [Release history](CHANGELOG.md)
 

README.zh-CN.md

@@ -10,7 +10,7 @@
 被阻断，还是走错了链路。
 
 `ipcheck` 是一个零依赖 Bash CLI，检测 AI 编程客户端实际使用的服务路径，
-提供可达性、TTFB 中位数/P95、抖动、参考上下行速度和直白的开发适配分。
+提供可达性、TTFB 中位数/P95、抖动、限量代理链路传输样本和直白的开发适配分。
 无需登录 Codex 或 Claude；它不会读取 API Key、发送 Prompt，也不会产生
 模型调用费用。
 
@@ -34,7 +34,7 @@ brew upgrade ipcheck
 
 ```bash
 mkdir -p "$HOME/.local/bin"
-curl -fsSL https://raw.githubusercontent.com/jacklv-coder/ipcheck/v0.8.2/bin/ipcheck \
+curl -fsSL https://raw.githubusercontent.com/jacklv-coder/ipcheck/v0.9.0/bin/ipcheck \
   -o "$HOME/.local/bin/ipcheck"
 chmod +x "$HOME/.local/bin/ipcheck"
 ```
@@ -49,7 +49,7 @@ chmod +x "$HOME/.local/bin/ipcheck"
 | --- | --- |
 | Speedtest | 到附近测速服务器的峰值带宽 |
 | `ping` / `curl` | 单个地址的基础连通性 |
-| `ipcheck` | Codex/Claude 实际路径、代理、TTFB、P95、抖动、失败率、带宽和开发适配分 |
+| `ipcheck` | Codex/Claude 实际路径、代理、TTFB、P95、抖动、失败率、限量参考传输和开发适配分 |
 
 即使带宽达到 500 Mbps，只要首字节需要五秒或代理链路不稳定，AI 编程体验
 仍然会很慢。`ipcheck` 直接测量这些影响交互节奏的瓶颈。
@@ -59,7 +59,8 @@ chmod +x "$HOME/.local/bin/ipcheck"
 - 分别检测每个客户端，避免一条健康链路掩盖另一条异常链路。
 - 兼容 OpenAI、Anthropic、LiteLLM 类网关，以及阿里云百炼/DashScope
   Anthropic 兼容入口。
-- 直接指出主要问题来自可达性、TTFB、P95、抖动、下载还是上传。
+- 直接指出主要问题来自可达性、TTFB、P95、抖动，还是 Cloudflare 参考传输
+  样本偏低。
 - 支持终端、Markdown 和稳定的版本化 JSON 报告。
 - 提供动态进度、窄终端适配，并能通过 `Ctrl+C` 干净退出，返回状态码 `130`。
 
@@ -81,14 +82,14 @@ Claude Code 的 Bedrock、Vertex AI、Foundry、Mantle 等 provider 原生协议
 | 命令 | 用途 |
 | --- | --- |
 | `ipcheck` | 自动识别客户端并执行标准检测 |
-| `ipcheck --quick` | 单次采样、缩短超时、不传输测速数据 |
+| `ipcheck --quick` | 单次采样、缩短超时、不运行参考传输 |
 | `ipcheck codex` / `ipcheck claude` | 只检测一个客户端 |
 | `ipcheck all` | 同时检测 Codex 和 Claude Code |
 | `ipcheck --explain-score` | 展开全部评分依据 |
 | `ipcheck --json` | 输出供自动化使用的结构化数据 |
 | `ipcheck --markdown` | 生成可分享的支持报告 |
 | `ipcheck --system` | 增加 macOS `networkQuality` 数据 |
-| `ipcheck --no-bandwidth` | 跳过限量 Cloudflare 上下行检测 |
+| `ipcheck --no-bandwidth` | 跳过限量 Cloudflare 参考传输 |
 
 运行 `ipcheck --help` 可以查看全部选项和环境变量。
 
@@ -108,8 +109,12 @@ API 路径；HTTP `407` 表示请求被代理拦截。
 TTFB 来自无凭据协议请求，反映 DNS、代理、TLS、网络与网关入口耗时；它不
 包含认证后的模型生成时间，也不是模型首个 Token 的到达时间。
 
-0–100 分是透明的规则评分，不代表用户百分位，并且服务可达性和交互延迟
-比峰值带宽更重要。精确计算方式见[评分与结论规则](docs/scoring.zh-CN.md)。
+Cloudflare 样本反映的是当前代理/网络到 Cloudflare 的小文件传输表现，不是
+峰值宽带测速，也不代表 OpenAI、Anthropic、GitHub 或 npm 的吞吐。
+
+0–100 分是透明的规则评分，不代表用户百分位。实际 AI 服务路径提供全部正向
+分数，Cloudflare 样本偏低时只会小幅扣分。精确计算方式见
+[评分与结论规则](docs/scoring.zh-CN.md)。
 
 ## Claude Code 网关示例
 
@@ -141,7 +146,7 @@ TTFB 来自无凭据协议请求，反映 DNS、代理、TLS、网络与网关
 ## 详细文档
 
 - [评分与结论规则](docs/scoring.zh-CN.md)
-- [服务路径、代理与带宽](docs/network.zh-CN.md)
+- [服务路径、代理与参考传输](docs/network.zh-CN.md)
 - [报告、自动化与退出码](docs/reporting.zh-CN.md)
 - [版本历史](CHANGELOG.md)