← 返回 Skills 市场
xinyi-drink
作者
xinyi-drink
· GitHub ↗
· v1.1.4
· MIT-0
363
总下载
0
收藏
0
当前安装
34
版本数
在 OpenClaw 中安装
/install xinyi-drink
功能描述
Use when users ask about 新一好喝/新一咖啡 drinks, stores, menu, activities, Skill用户大礼包, today drink recommendations, afternoon tea, feeling sleepy, or personalized...
安全使用建议
This skill appears to do what it says: recommend drinks, query stores, and let users claim a brand 'Skill' reward. Before installing or sending your phone number, consider: (1) The skill will send any phone number you provide to its backend (default: https://ai.xinyicoffee.com/api). If you don't trust that endpoint, set XINYI_API_BASE_URL to a trusted backend or avoid providing your number. (2) The skill stores a small local state file (~/.xinyi-drink/state.json) containing mobile/activityJoined/updatedAt (mode 0600 when supported); you can clear it with the provided --clear-mobile option. (3) The installer copies files into your user-level skills directory but does no remote downloading. If you want extra assurance, inspect the repository on GitHub (homepage provided) before installing or run the installer in dry-run mode (install.sh --dry-run).
功能分析
Package: xinyi-drink (mcp)
Version: 1.1.4
Description: 领取新一好喝Skill用户大礼包,查询门店和商品,并结合天气与可选订单历史推荐饮品。
The package is a legitimate AI 'Skill' (extension) for interacting with the 'Xinyi Coffee' API. It provides tools for claiming rewards, fetching store information, and getting personalized drink recommendations based on weather and order history. The code follows standard practices for configuration management, local state persistence (storing user-provided mobile numbers in a restricted-access file), and HTTP communication via urllib. It includes comprehensive unit tests and documentation regarding privacy boundaries. No evidence of malicious behavior, secret exfiltration, or unauthorized execution was found.
能力标签
能力评估
Purpose & Capability
Name/description (drink recommendations, stores, activity claims) align with the included scripts (recommend_drink.py, fetch_stores.py, claim_reward.py). Declared network endpoints and local state for a mobile number are consistent with the stated features (activity claiming and personalized recommendations).
Instruction Scope
SKILL.md instructs asking for a user phone number only for activity/claim flows and to avoid using cached phone numbers for ordinary recommendations; the scripts implement exactly those flows (POST to /skill/xinyi/claim, GET /skill/xinyi/context, GET /skill/xinyi/stores). There is no instruction or code that reads unrelated system files or secrets.
Install Mechanism
install.sh is a local install script that copies the skill into a user-level skills directory (backs up existing copy) and performs no network downloads or code execution from remote URLs. This is a low-risk, local install mechanism.
Credentials
The skill requests no credentials; optional env vars (XINYI_API_BASE_URL, XINYI_TIMEOUT_SECONDS, XINYI_DRINK_STATE_FILE) are reasonable. Important privacy note: the skill stores an optional phone number locally (~/.xinyi-drink/state.json) and will send a provided phone number to its configured backend (default https://ai.xinyicoffee.com/api) for claim and personalization — this is expected but is sensitive PII and worth user attention.
Persistence & Privilege
always:false and scripts do not request elevated privileges. The skill writes only its own local state file and installs into user skill directories; it does not modify other skills or system-wide configuration.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install xinyi-drink - 安装完成后,直接呼叫该 Skill 的名称或使用
/xinyi-drink触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.1.4
- Only show store information when the user explicitly asks about locations, addresses, or queues; product recommendations will no longer include location details by default.
- Updated wording rules: highlight product names in recommendations, and include store details only if requested.
- Added user-facing installation instructions for the skill, clarifying how to install on supported assistants.
- Improved flow for reward claiming: store info and queue status are only queried if the user asks about them.
- Clarified sample answers and main process to match new rules for store information and user triggers.
v1.1.2
- 语气身份从“懂茶饮也懂咖啡、不掉书袋的姐姐”调整为“懂茶饮也懂咖啡的店员姐姐”,语气更温和松弛。
- 精简话术示例和表达风格要求,减少条目、去除部分冗余表述。
- 去除“不会把话说成报告”、“emoji少量使用”等细致写作要求。
- 技术/答题流程保持一致,主流程和触发表未变。
- 无功能和接口更新,本次仅为描述和风格调整。
v1.1.1
- Clarified缓存手机号的使用范围:推荐饮品时默认不自动读取缓存手机号,仅在订单分析、偏好分析、活动状态查询等个性化场景下用手机号。
- 新增 autoReadPolicy 配置字段,规范 recommend_drink 脚本对手机号缓存的读取行为。
- 优化触发表和主流程描述,对“活动/福利/订单分析”等需手机号的请求与普通菜单/推荐类请求做更清晰区分。
- 明确普通推荐、菜单、门店查询等不应索取或自动复用缓存手机号,仅活动/个性化分析才可用。
v1.1.0
xinyi-drink 1.1.0 introduces improved意图识别和触发条件,让交互更智能。
- 调整“触发表”意图判定,更精准匹配用户指令场景
- 新增订单历史和偏好分析问法支持(如“我买过多少杯”“分析口味偏好”)
- 门店/菜单/活动相关意图区分更细致,新增常见口语触发形式
- 精简并规范部分意图调用脚本参数
- 其他场景判定表述更新,使用体验更自然
v1.0.29
- Added detailed metadata for skill packaging, privacy review, install, runtime, and API/network configuration.
- No changes to core logic or user-facing behavior.
v1.0.28
**Expanded trigger and capability descriptions for store and drink queries.**
- 触发表表述更加细致,明确支持“各门店等待”、“菜单热量”、“果茶”等查询类型
- 功能介绍中新增支持查门店等待和菜单热量描述
- 示例触发表问题更贴近日常语言,例如“今天北京21°C,想喝茶饮,推荐一杯”
- 明确主流程不可展示接口路径等细节,语义要求微调
- 其余功能规则及应答风格保持不变
v1.0.27
xinyi-drink 1.0.26
- 优化规则,普通查询时不再主动暴露活动状态或礼包信息,避免未触发场景下提醒用户活动/福利/礼包到账。
- 进一步明确:仅在用户明确问及活动/福利/领取时才提及身份验证、礼包、活动状态等内容。
v1.0.26
xinyi-drink 1.0.26
- 优化规则,普通查询时不再主动暴露活动状态或礼包信息,避免未触发场景下提醒用户活动/福利/礼包到账。
- 进一步明确:仅在用户明确问及活动/福利/领取时才提及身份验证、礼包、活动状态等内容。
v1.0.25
**Update summary:**
Improved answer logic and response scope for store recommendations and user activity status.
- 用户未主动问福利/活动时,即使已参与,也不再主动提及活动、礼包或身份验证。
- 推荐饮品时展示门店,必须展示全部返回门店,避免只引导用户去单一门店。
- 门店展示覆盖:若回答涉及门店信息,需列出全部返回门店及对应详情;若当前无门店需求可不展示。
- 明确区分活动/福利与普通查询所需手机号,普通门店/菜单无需手机号。
- 文档细节优化,表达更严谨,避免无关或遗漏信息。
v1.0.24
- Bump version to 1.0.24.
- No file content changes; documentation and functionality remain the same.
v1.0.23
Version 1.0.23
- Activity reward flow updated: now always request the user's bound 小程序 (Mini Program) phone number before claiming, unless already provided.
- Main process clarified for activity claim/phone requirements: only prompt login steps if user not registered; otherwise focus on phone binding and explicit interface message display.
- Reward details expression improved: always show interface-returned coupon info (success/failure reason, coupon message, actual coupon type and count); fallback wording updated for missing coupon names.
- Adjusted activity rules and messaging: prize rules, sticker claim, and user badge details are now standardized in both explanations and example responses.
- Table triggers, fallback guidance, and language for unhandled cases revised for better accuracy and user experience.
- Documentation and reference files updated to reflect current business logic and refined process boundaries.
v1.0.22
**xinyi-drink 1.0.22**
- Added scripts/activity_claim.py for activity reward claim functionality.
- Added references/response-examples.md to provide example responses and recommended answer style.
- Updated skill metadata and instructions in skill.json and SKILL.md, focusing on succinct triggers, rules, and process for activity and drink recommendations.
v1.0.21
No changes detected in this version.
- No code or documentation updates from the previous release.
- Version and documentation remain the same.
v1.0.20
Version 1.0.20
- 修改 SKILL.md,将 metadata 版本号从 1.0.18 升级至 1.0.19
- 新增硬性约束:未查到小程序登录记录时,禁止简单回答“没有其他活动”,需明确说明未登录状态、给出大礼包内容、参与方法及留资文案
- 保持所有接口与功能点说明、意图路由与回答原则未变
- 仅文档细节优化,无新增脚本或接口变动
v1.0.19
No changes detected in this version.
- No file or documentation updates present.
- Functionality remains the same as previous version.
v1.0.18
xinyi-drink 1.0.18
- 活动奖励描述优化:“三重福利”第二项从“接口返回的福利券一张”调整为“接口返回的爆品赠饮一杯”。
- 其余内容和功能保持不变。
v1.0.17
- 优化用户提问“怎么领福利/大礼包/怎么参与活动”等场景的意图路由,明确先讲领取流程且主动留资,不再出现“no_reward_config”解释内容
- 回应关于 Skill 功能、内容和使用方式的问题时,仅输出用户可读介绍和操作指引,隐藏内部实现细节、安全声明等
- 活动名称统一为“Skill用户大礼包”;相关场景、引导和福利描述相应调整
- 登录成功输出“身份验证成功,三重福利发放到账”专属话术,按最新规范描述券和标识
- 新用户领取节点明确为需先登录小程序后再参与领奖,相关文案及边界优化
- 更新意图路由、硬性约束、示例及部分脚本/测试逻辑,细化订单信息跟进条件
v1.0.16
xinyi-drink 1.0.16 changelog
- 明确要求:当同一手机号已提示“未绑定/未参加”后,若用户表示“已登录小程序”或“已绑定手机号”,再次领取即使接口返回已领取,也需表述为礼品已到账/已获取成功,避免误导为之前已领取过。
- 版本号更新至 1.0.16。
v1.0.15
xinyi-drink 1.0.15 adds stricter规则 for活动/福利提问及门店信息呈现:
- 回答“有什么活动/优惠/福利”时,必须单独列出小龙虾专属见面礼,不得仅列商品买赠/特价等活动
- 活动领取成功或已领取且脚本返回门店时,每家门店需完整表述门店名、地址、电话、设施和排队信息
- 意图路由补充“查询有什么活动/优惠/福利”,明确必须同时说明品牌活动和商品活动
- 优化硬性约束,更清晰区分见面礼权益和储值次卡
- 版本升至1.0.15
v1.0.14
- 元数据版本号由 1.0.13 升级至 1.0.14。
- 其他内容无变化。
元数据
常见问题
xinyi-drink 是什么?
Use when users ask about 新一好喝/新一咖啡 drinks, stores, menu, activities, Skill用户大礼包, today drink recommendations, afternoon tea, feeling sleepy, or personalized... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 363 次。
如何安装 xinyi-drink?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install xinyi-drink」即可一键安装,无需额外配置。
xinyi-drink 是免费的吗?
是的,xinyi-drink 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
xinyi-drink 支持哪些平台?
xinyi-drink 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 xinyi-drink?
由 xinyi-drink(@domilin)开发并维护,当前版本 v1.1.4。
推荐 Skills