# coding-proxy

> A High-Availability, Transparent, and Smart Multi-Vendor Proxy for Claude Code. Support Claude Plans, GitHub Copilot, Google Antigravity, ZAI/GLM, MiniMax, Qwen, Xiaomi, Kimi, Doubao...

- **URL**: https://www.freshcrate.ai/projects/coding-proxy
- **Author**: ThreeFish-AI
- **Category**: Infrastructure
- **Latest version**: `v0.5.1a7` (2026-06-07)
- **License**: Apache-2.0
- **Source**: https://github.com/ThreeFish-AI/coding-proxy
- **Language**: Python
- **GitHub**: 13 stars, 1 forks
- **Registry**: github
- **Tags**: `antigravity`, `claude-code`, `copilot`, `doubao`, `glm`, `kimi`, `llm-agent`, `minimax`, `python`

## Description

A High-Availability, Transparent, and Smart Multi-Vendor Proxy for Claude Code. Support Claude Plans, GitHub Copilot, Google Antigravity, ZAI/GLM, MiniMax, Qwen, Xiaomi, Kimi, Doubao...

## Recent releases

| Version | Date | Urgency | Changes |
| --- | --- | --- | --- |
| `v0.5.1a7` | 2026-06-07 | High | - feat(dashboard): Model Calling 实时监控扩展至全 vendor，修复排队队列长度不可见 (#253) - fix(session-title): 剥离 标签前缀并放大标题存储长度至 600 字符 (#257) - feat(session-title): 多层级回退标题提取与延迟补写，消除 Dashboard 空标题 (#258) - feat(session-routing): 新增基于 Session 标题前缀的可配置 Vendor 自动绑定 (#260) |
| `v0.5.1a4` | 2026-05-31 | High | - feat(dashboard): Model Calling 实时监控扩展至全 vendor，修复排队队列长度不可见 (#253) - fix(session-title): 剥离 标签前缀并放大标题存储长度至 600 字符 (#257) |
| `v0.4.1a6` | 2026-05-23 | High | - feat(zhipu): 为 429 Rate Limit 添加指数退避重试挽回机制 (#242) - fix(antigravity): 修复 v1internal 模式检测逻辑并新增 E2E 测试; (#234) - fix(routes): 修复 count_tokens 路由对 target_vendor.name 的错误属性访问; (#235) - fix(vendor-channels): 修复 zhipu→anthropic 通道 tool_use/tool_result 配对漏洞; (#236) - fix(native-api): 修复 Gemini :verb 路径中 %3A URL 编码导致上游 400 的兼容问题; (#237) - fix(zhipu): 诊断首选 tier 语义拒绝降级问题，增强可观测性并提取跨供应商清洗共享函数 (#243) |
| `0.4.1a4` | 2026-05-16 | High | - fix(antigravity): 修复 v1internal 模式检测逻辑并新增 E2E 测试; (#234) - fix(routes): 修复 count_tokens 路由对 target_vendor.name 的错误属性访问; (#235) - fix(vendor-channels): 修复 zhipu→anthropic 通道 tool_use/tool_result 配对漏洞; (#236) - fix(native-api): 修复 Gemini :verb 路径中 %3A URL 编码导致上游 400 的兼容问题; (#237) |
| `v0.4.1a2` | 2026-05-13 | High | - fix(antigravity): 修复 v1internal 模式检测逻辑并新增 E2E 测试; (#234) - fix(routes): 修复 count_tokens 路由对 target_vendor.name 的错误属性访问; (#235) - fix(vendor-channels): 修复 zhipu→anthropic 通道 tool_use/tool_result 配对漏洞; (#236) - fix(native-api): 修复 Gemini :verb 路径中 %3A URL 编码导致上游 400 的兼容问题; (#237) |
| `v0.4.0` | 2026-05-01 | High | > [!IMPORTANT] > > **🚀 Session 级专属路由策略！** > > 给每个 Session 指定专属的 vendor，动态调节不同 vendors 间的 LLM 流量。  ![session](assets/session-v0.4.0.png)  ### ✨ 核心亮点  - feat(session-policy): 新增 Session 级专属路由策略 (#219) - feat(dashboard): 新增会话活动面板 (#222)  ### 🔧 更多特性  - refactor(logging): 移除已被 ModelCall 汇总行覆盖的冗余 DEBUG 日志 (#203) - style(dashboard): 加宽图表 tooltip 令模型名称与用量值单行显示 (#211) - fix(usage-parser): 补充 OpenAI/Gemini SSE 流式分支的 model_served 提取 (#214) - fix(usage-parser): 兼容 SSE chunk 中 usage 字段为 n |
| `v0.3.1a1` | 2026-04-26 | High | - refactor(logging): 移除已被 ModelCall 汇总行覆盖的冗余 DEBUG 日志; |
| `v0.3.0` | 2026-04-20 | High | > [!IMPORTANT] > > **🚀 OpenAI、Anthropic、Gemini 原生 API 进驻 Coding Proxy！** > > 服务对象不在局限于 Claude Code，凡兼容 OpenAI、Anthropic、Gemini 三巨头 API 协议的客户端，出口 LLM 流量可统一收敛到 Coding Proxy。  ### ✨ 核心亮点  - feat(native-api): 新增 `/api/{openai,gemini,anthropic}/**` 原生 LLM API 全量 catch-all 透传通道； - feat(dashboard): 新增实时 Web Dashboard 页面，聚合展示流量与用量统计; - feat(usage): `usage` 区分 Claude Code 场景（`'cc'`）与原生 API 场景（`'api'`）； - refactor(vendor-channels): 将供应商转换通道从目标专属重构为源→目标绑定模型; - docs(user-guide): 补充 POST /v1/mes |
| `v0.3.0a3` | 2026-04-20 | High | - feat(native-api): 新增 `/api/{openai,gemini,anthropic}/**` 原生 LLM API 全量 catch-all 透传通道； - feat(usage): `usage_log` 新增 `client_category` / `operation` / `endpoint` / `extra_usage_json` 四列，区分 Claude Code 场景（`'cc'`）与原生 API 场景（`'api'`），承载规范化操作名与非规范 token 字段（reasoning / audio / thoughts / server_tool_use 等）； - refactor(vendor-channels): 将供应商转换通道从目标专属重构为源→目标绑定模型; - refactor(thinking-strip): 条件化 thinking block 剥离，仅跨供应商场景触发; - style(dashboard): 缩小「今日费用估算」卡片费用数字字体使其单行显示; |
| `v0.3.0a2` | 2026-04-19 | High | - feat(native-api): 新增 `/api/{openai,gemini,anthropic}/**` 原生 LLM API 全量 catch-all 透传通道； - feat(usage): `usage_log` 新增 `client_category` / `operation` / `endpoint` / `extra_usage_json` 四列，区分 Claude Code 场景（`'cc'`）与原生 API 场景（`'api'`），承载规范化操作名与非规范 token 字段（reasoning / audio / thoughts / server_tool_use 等）； - refactor(vendor-channels): 将供应商转换通道从目标专属重构为源→目标绑定模型; - refactor(thinking-strip): 条件化 thinking block 剥离，仅跨供应商场景触发; - style(dashboard): 缩小「今日费用估算」卡片费用数字字体使其单行显示; |

## Dependency audit

- **Score**: 77/100
- **Total deps**: 8
- **Resolved**: 3
- **Unresolved**: 5
- **License conflicts**: 0
- **Warnings**: 5
- **Scanned**: 2026-06-01

## Citation

- HTML: https://www.freshcrate.ai/projects/coding-proxy
- Markdown: https://www.freshcrate.ai/projects/coding-proxy.md
- Dependencies JSON: https://www.freshcrate.ai/api/projects/coding-proxy/deps

_Generated by freshcrate.ai. Indexes github releases for AI-agent ecosystem packages._