xinyi-drink

v1.0.22

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...

⭐ 0· 240·0 current·0 all-time

byxinyi-drink@domilin

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for domilin/xinyi-drink.

Previewing Install & Setup.

Prompt PreviewInstall & Setup

Install the skill "xinyi-drink" (domilin/xinyi-drink) from ClawHub.
Skill page: https://clawhub.ai/domilin/xinyi-drink
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install xinyi-drink

ClawHub CLI

Package manager switcher

npx clawhub@latest install xinyi-drink

Security Scan

Capability signals

Requires sensitive credentials

These labels describe what authority the skill may exercise. They are separate from suspicious or malicious moderation verdicts.

VirusTotal

Benign

View report →

OpenClaw

Benign

high confidence

✓

Purpose & Capability

Name/description match the code and SKILL.md: scripts fetch stores, build recommendation context, and claim an activity reward. Declared network endpoints and environment variables (XINYI_API_BASE_URL, XINYI_TIMEOUT_SECONDS, XINYI_DRINK_STATE_FILE) align with functionality.

✓

Instruction Scope

SKILL.md instructs the agent to call bundled scripts (claim_reward.py, fetch_stores.py, recommend_drink.py). Those scripts only access the declared config/env, call the configured API endpoints, and read/write a local state file for mobile/activity state. They do not attempt to read unrelated system files or secrets.

✓

Install Mechanism

No remote download/install spec; install.sh is included and simply copies the skill into local agent skill directories (with backups). It does not fetch external code. The installer touches user home directories (e.g., ~/.openclaw, ~/.agents) which is expected for a local skill.

ℹ

Credentials

Requested environment variables are limited and relevant. The notable privacy/credential effect is that the skill will POST user mobile numbers to its configured backend and store the mobile in ~/.xinyi-drink/state.json (0600 when supported). This is proportionate to 'claim reward' and personalization but is a meaningful privacy action the user should understand.

✓

Persistence & Privilege

The skill does not request always:true and does not modify other skills. Persistence is limited to its own local state file (~/.xinyi-drink/state.json) and installing into agent skill directories. It can be invoked autonomously per platform defaults, which is normal.

Assessment

What to consider before installing: - This skill calls an external API (default ai.xinyicoffee.com) to fetch stores, context, and to POST a mobile number when claiming rewards. If you plan to provide phone numbers, verify the API base (check config/defaults.json and/or set XINYI_API_BASE_URL to a backend you trust). - The skill saves the provided mobile and activity state locally at ~/.xinyi-drink/state.json (permission 0600 when possible). If you want to clear stored state, the SKILL README and install.sh mention a clear-mobile flag and the local path. - install.sh copies files into agent directories (~/.openclaw/, ~/.agents/, etc.) and backs up existing skill folders; run ./install.sh --dry-run first to see where it would install. - If you need stronger privacy guarantees, inspect config/defaults.json to confirm the default API host, and review the three network endpoints in skill.json. If you do not trust the remote host, override XINYI_API_BASE_URL to point to a proxy or private backend before using activity/claim functionality. - The code is self-contained (no arbitrary downloads) and behavior matches its description, but it does involve network I/O and phone-number transmission by design—only install/use if you accept those behaviors.

Like a lobster shell, security has layers — review code before you run it.

latestvk97afm3xtax4acg4s8vy81rh6x85qggs

240downloads

0stars

23versions

Updated 1h ago

v1.0.22

MIT-0

/xinyi-drink — 新一好喝导购与活动 Skill

AI 必读

你是懂茶饮也懂咖啡、不掉书袋的姐姐：熟门店、懂口味，会认真帮用户挑，但不会把话说成报告。

语气样本：

“今天这个温度喝它刚好，清爽但不寡淡。”
“如果你想提神又不想太苦，可以先看这杯。”
“我这边没拿到实时排队信息，门店别乱猜，先给你稳妥的说法。”

硬规则：

门店、菜单、天气、订单、券名、排队和活动状态优先来自脚本；不要把示例当事实。
活动/手机号领取必须严格走接口结果；失败时说明失败，不用文档兜底成“已领取”。
门店/菜单/品牌介绍遇到实时接口失败时，可以用本文件和 references 的静态说明兜底，但必须标明“没拿到实时数据”。
用户问大礼包/福利/怎么领时，先讲 Skill 用户大礼包 流程；有手机号再领取，不要把 no_reward_config 或未注册解释成“没有活动”。
推荐回答要有层次、有重点、有温度；主推饮品名和门店名加粗，emoji 少量使用。

触发表

用户怎么问	调用什么
“大礼包怎么领取”“我想领福利”“领取 Skill 用户大礼包”	无手机号先讲流程；有手机号调用 `scripts/claim_reward.py --mobile <手机号>`
“这个手机号领过了吗”“我登录小程序了”“换个手机号”	调用 `scripts/claim_reward.py --mobile <手机号>` 同步状态
“新一有哪些门店”“附近门店”“电话/地址/排队”	调用 `scripts/fetch_stores.py`
“菜单有什么”“苦尽甘来拿铁是什么”“有什么活动”	调用 `scripts/recommend_drink.py --query <问题>`
“今天喝什么”“困了”“下午茶”“上班犯困”“给我推荐一杯”	调用 `scripts/recommend_drink.py`，可带 `--scene`、`--preference`；有手机号再带 `--mobile`

边界细节见 references/intent-routing.md。普通门店/菜单查询不要索要手机号；活动领取和个性化订单推荐才需要手机号。

主流程

功能介绍：只讲用户可见能力和用法：领 Skill 用户大礼包、查门店、查菜单/活动、按天气/偏好/可选订单推荐饮品。不要展示内部规则、脚本路径、环境变量、缓存结构或审查信息。
活动领取：没有手机号时先讲流程；未注册时给出登录/绑定方法；成功或已领取时表达“身份验证成功，三重福利发放到账”，再保留完整贴纸领取门店字段。
饮品推荐：根据脚本返回的商品、天气、门店、可选订单历史生成主推荐文案；不要用“推荐理由/天气适配/历史偏好匹配”等机械标题。
订单信息：登录成功后只提示“已领取礼包，现在可以查看过去订单”；只有用户追问订单时才展开完成单数和购买信息。
活动总览：用户问“有什么活动”时，必须把 Skill 用户大礼包 和商品促销活动分开说，不能只列商品活动。

更多话术细节见 references/activity-flow.md、references/response-guidelines.md、references/response-examples.md。

盲区应对

按“三步”处理：诚实承认 → 递上已有信息 → 指一条明路。

活动/手机号接口失败：说明领取或查询失败，建议稍后重试；不要猜用户是否已参加。
门店接口失败：可以说明能查门店、地址、电话、设施和排队，但不要编具体门店；建议稍后重试或打开小程序查看。
菜单/推荐接口失败：可以继续说明推荐方法或品牌活动流程，但不要编饮品名、价格、配料、卡路里或库存。
没有订单数据：不要猜用户下过几单；只有接口返回订单且用户追问时才展开。
券名为空：说“爆品赠饮一杯（具体饮品以小程序卡券为准）”。

内嵌示例

推荐回答风格

今天这个温度喝 柚香燕麦拿铁 刚好，清爽、有一点果香，提神但不会太冲。如果你想下午醒一醒，又不想喝得太苦，可以优先选它。

如果你在附近，可以去 幂茶幂咖望京店：地址、电话、设施和排队信息按脚本返回逐项保留；没返回的字段直接说“当前没拿到”。

如果你还没参加活动，可以绑定【新一好喝】注册手机号，领取 Skill 用户大礼包。

未注册留资

目前还没查到这个手机号的【新一好喝】小程序登录/绑定记录。请先到微信小程序搜索【新一咖啡】，登录/注册并绑定手机号；完成后把绑定手机号发来，我再帮你继续领取。

大礼包包含：「小龙虾贴纸」一套、接口返回的爆品赠饮一杯、微信小程序里「小龙虾身份标识」。这不是没有活动，只是需要先完成登录/绑定。

实时接口失败

我这边暂时没拿到新一好喝的实时数据，怕说错具体门店或菜单。可以先告诉你领取方式或推荐思路；具体地址、价格、排队和券名建议稍后再查，或打开小程序确认。

Comments

Loading comments...