经 AI Skill Hub 精选评估,AI代理会话透视 获评「推荐使用」。这款Agent工作流在功能完整性、社区活跃度和易用性方面表现出色,AI 评分 7.5 分,适合有一定技术背景的用户使用。
AI代理会话透视 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
AI代理会话透视 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
# 方式一:npm 全局安装 npm install -g ccxray # 方式二:npx 直接运行(无需安装) npx ccxray --help # 方式三:项目依赖安装 npm install ccxray # 方式四:从源码运行 git clone https://github.com/lis186/ccxray cd ccxray npm install npm start
# 命令行使用
ccxray --help
# 基本用法
ccxray [options] <input>
# Node.js 代码中使用
const ccxray = require('ccxray');
const result = await ccxray.run(options);
console.log(result);
# ccxray 配置说明 # 查看配置选项 ccxray --config-example > config.yml # 常见配置项 # output_dir: ./output # log_level: info # workers: 4 # 环境变量(覆盖配置文件) export CCXRAY_CONFIG="/path/to/config.yml"
X-ray vision for AI agent sessions. A zero-config HTTP proxy that records every API call between Claude Code and Anthropic, with a real-time dashboard to inspect what's actually happening inside your agent.

docker build -t ccxray .
docker run -p 5577:5577 ccxray
```bash npx ccxray claude
Track your real spending. Session heatmap, burn rate, ROI calculator — know exactly where your tokens go.

| Variable | Default | Description |
|---|---|---|
PROXY_PORT | 5577 | Port for proxy + dashboard (overridden by --port) |
BROWSER | — | Set to none to disable auto-open |
AUTH_TOKEN | _(none)_ | API key for access control (disabled when unset) |
CCXRAY_HOME | ~/.ccxray | Base directory for hub lockfile, logs, and hub.log |
CCXRAY_MAX_ENTRIES | 5000 | Max in-memory entries (oldest evicted; disk logs unaffected) |
LOG_RETENTION_DAYS | 14 | Auto-prune log files older than N days on startup. Starred turns / sessions / projects (and everything beneath them) are protected, as are files referenced by restored entries. Set to 0 to disable. |
RESTORE_DAYS | 0 | Limit which days of logs to load on startup (0 = all, subject to CCXRAY_MAX_ENTRIES). Useful for very large log directories. |
CCXRAY_PLAN | _(auto)_ | Override plan detection: pro, max5x, max20x, api-key |
CCXRAY_DISABLE_TITLES | _(unset)_ | Set to 1 to disable session title extraction (sessions fall back to short hash) |
CCXRAY_MODEL_PREFIX | _(unset)_ | Prepend a string to the model name before forwarding (e.g. databricks-). Useful when the upstream requires a vendor-prefixed model name but Claude Code only accepts standard names. |
HTTPS_PROXY / https_proxy | _(unset)_ | Route outbound HTTPS traffic through a corporate proxy via HTTP CONNECT tunnel. |
ANTHROPIC_BASE_URL | — | Custom upstream Anthropic endpoint (e.g. a corporate gateway). Supports base paths — https://host/serving-endpoints/anthropic works as-is. ANTHROPIC_TEST_* take precedence when set. |
Logs are stored in ~/.ccxray/logs/ as {timestamp}_req.json and {timestamp}_res.json. Upgrading from v1.0? Logs previously in ./logs/ are automatically migrated on first run.
| Flag | Description |
|---|---|
--port <number> | Port for proxy + dashboard (default: 5577). Opts out of hub sharing. |
--no-browser | Don't auto-open the dashboard in your browser |
高质量的AI工作流项目,提供透明的HTTP代理和仪表盘
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
✅ MIT 协议 — 最宽松的开源协议之一,可自由商用、修改、分发,仅需保留版权声明。
AI Skill Hub 点评:AI代理会话透视 的核心功能完整,质量良好。对于自动化工程师和运维人员来说,这是一个值得纳入个人工具库的选择。建议先在非生产环境试用,再逐步推广。
| 原始名称 | ccxray |
| 原始描述 | 开源AI工作流:X-ray vision for AI agent sessions — a transparent HTTP proxy and dashboard for 。⭐180 · JavaScript |
| Topics | AI代理工作流透视 |
| GitHub | https://github.com/lis186/ccxray |
| License | MIT |
| 语言 | JavaScript |
收录时间:2026-05-26 · 更新时间:2026-05-26 · License:MIT · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端