← Back to Skills Marketplace
Lesson
by
yequanzheng
· GitHub ↗
· v0.1.1
· MIT-0
217
Downloads
0
Stars
8
Active Installs
2
Versions
Install in OpenClaw
/install lesson
Description
生成今日菜单推荐。根据用户手头的食材、口味偏好、人数和烹饪时间, 从内置食材库和菜谱库中智能匹配,输出完整的菜单方案和分步烹饪指南。 当用户提到"今天吃什么""做什么菜""菜单""菜谱推荐""晚饭吃啥"等意图时触发。
README (SKILL.md)
今日菜单生成器
根据用户现有食材、口味偏好、就餐人数和可用时间,生成个性化菜单推荐和烹饪指南。
工作流程
第一步:了解用户需求
通过简短对话收集以下信息(缺失项用合理默认值):
| 信息项 | 默认值 | 说明 |
|---|---|---|
| 手头食材 | — | 必问,至少确认 1~3 种主要食材 |
| 就餐人数 | 2 人 | 影响份量建议 |
| 口味偏好 | 家常 | 家常 / 清淡减脂 / 川湘重口 / 快手10分钟 |
| 可用时间 | 60 分钟 | 影响菜品复杂度 |
| 忌口/过敏 | 无 | 如有则排除相关食材 |
| 菜品数量 | 2~4 道 | 根据人数自动建议 |
对话原则:
- 不要一次问太多问题,最核心的是「你手头有什么食材?」
- 如果用户只说了食材,其他用默认值直接推荐
- 语气轻松友好,像朋友聊天一样
第二步:匹配菜谱
- 读取
references/ingredients.md获取完整食材分类 - 读取
references/recipes.md获取菜谱库 - 根据用户提供的食材,从菜谱库中匹配可做的菜品
- 考虑荤素搭配、营养均衡、烹饪难度
匹配规则:
- 主料必须匹配(用户手头有的食材)
- 常见调料(盐、酱油、醋、糖、料酒、蚝油等)默认家里有,不需要确认
- 优先推荐主料完全匹配的菜,其次推荐只差 1~2 种配料的菜
- 如果用户食材能做的菜不够,可以建议额外采购少量食材
第三步:输出菜单
使用以下格式输出:
## 🍽️ 今日菜单推荐
**就餐人数:** X 人 | **预计用时:** X 分钟 | **风格:** XXX
---
### 1. 菜名 ⭐
> 一句话点评这道菜的特点
**食材:**
- 主料:XXX
- 配料:XXX
- 调料:XXX
**做法:**
1. 第一步(附时间和火候)
2. 第二步
3. ...
**小贴士:** 实用烹饪技巧
---
### 2. 菜名
(同上格式)
---
## 📋 备菜清单
把所有菜需要的食材汇总,方便一次性准备:
| 分类 | 食材 | 用量 | 处理方式 |
|------|------|------|----------|
| 肉类 | ... | ... | ... |
| 蔬菜 | ... | ... | ... |
## ⏱️ 烹饪时间线
建议的做菜顺序和并行操作:
1. 先做 XX(耗时最长的先开始)
2. 在等待的同时准备 XX
3. 最后快炒 XX
第四步:后续互动
菜单输出后,主动询问:
- 「要不要换掉某道菜?」
- 「要不要看更详细的某道菜做法?」
- 「还想加个汤/凉菜吗?」
口味风格指南
家常风格(默认)
- 红烧、清炒、炖煮为主
- 调味适中,老少皆宜
- 代表菜:红烧肉、番茄炒蛋、清炒时蔬
清淡减脂
- 少油少盐,以蒸、煮、白灼为主
- 多蔬菜,蛋白质优先选鸡胸肉/鱼/虾/豆腐
- 减少碳水,避免油炸
- 标注估算热量
快手 10 分钟
- 只推荐炒、拌、煮等快速烹饪方式
- 食材处理简单(切丝切片为主)
- 每道菜控制在 10 分钟以内
川湘重口味
- 麻辣鲜香,重油重味
- 豆瓣酱、辣椒、花椒、剁椒等调料
- 代表菜:麻婆豆腐、辣子鸡、酸辣土豆丝
重要原则
- 实用优先:做法描述要具体到火候、时间、用量,不要含糊
- 荤素搭配:2 道菜以上时确保有荤有素
- 难度递进:多道菜时从简到难排列
- 时间合理:给出的总用时要考虑并行操作
- 食材不浪费:尽量用完用户提到的食材
- 安全第一:涉及生食(如三文鱼)要提醒新鲜度要求
Usage Guidance
This skill appears coherent and low-risk: it uses local recipe and ingredient files and a local Python script to produce menus. Before installing, consider: (1) Confirm your agent/runtime will not execute additional downloads or run unreviewed code with elevated privileges—this skill's files look local and benign. (2) If you plan to allow the agent to execute the Python script, ensure the runtime environment is trusted (scripts can read files present in the skill bundle). (3) Check recipe text for allergy/diet accuracy if you have dietary restrictions. (4) If you need networked features (shopping links, external nutrition APIs), those would require additional permissions—this skill currently does not request them. Overall, nothing here appears disproportionate to a menu/recipe generator.
Capability Analysis
Type: OpenClaw Skill
Name: lesson
Version: 0.1.1
The skill bundle is a legitimate tool for generating personalized meal recommendations and cooking guides. The Python script (scripts/generate_menu.py) contains standard logic for ingredient matching and menu generation without any network calls, obfuscation, or dangerous system executions. The instructions in SKILL.md are strictly aligned with the culinary purpose and do not attempt to manipulate the agent into performing unauthorized actions or exfiltrating data.
Capability Assessment
Purpose & Capability
Name/description (daily menu / recipe recommendation) match the included assets: SKILL.md, a recipe DB, an ingredient DB, and a local Python script to generate menus. Nothing requested (no env vars, no binaries) is unrelated to the cooking purpose.
Instruction Scope
Runtime instructions explicitly limit actions to local data and user prompts (read references/ingredients.md and references/recipes.md, ask user about ingredients/preferences, generate formatted menu). There are no instructions to read arbitrary system files, environment variables, or to send data to external endpoints.
Install Mechanism
No install spec is provided; this is effectively instruction-only with bundled reference files and a script. No downloads, package installs, or external installers are used.
Credentials
The skill declares no required environment variables, credentials, or config paths. The Python script imports only standard libraries (argparse, json, sys, pathlib, typing) and contains no signs of accessing os.environ, network libraries, or credential stores.
Persistence & Privilege
always is false and the skill does not request persistent system privileges or to modify other skills or global agent settings. It appears to operate only when invoked.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install lesson - After installation, invoke the skill by name or use
/lesson - Provide required inputs per the skill's parameter spec and get structured output
Version History
v0.1.1
Initial release of the daily-menu skill for personalized meal recommendations.
- Generates daily menu suggestions based on available ingredients, taste preferences, number of diners, and cooking time.
- Pulls from built-in ingredient and recipe references to match suitable dishes.
- Outputs a full menu, ingredient prep list, step-by-step cooking guides, and a recommended cooking timeline.
- Interactive follow-up: suggests swapping dishes, adding soups/salads, or providing more detailed recipes.
- Friendly, conversational prompts with practical and clear instructions.
v0.1.0
Initial release with lesson extraction and storage functionality.
- Adds /lesson command to store key lessons from recent conversation context.
- Automatically extracts both technical (pitfall/fix) and decision principle (behavioral rule) layers.
- Verifies that both types of entries are retrievable via memory search.
- Reports a brief summary of stored lessons to the user.
- Prompts for clarification if no clear lesson is detected.
Metadata
Frequently Asked Questions
What is Lesson?
生成今日菜单推荐。根据用户手头的食材、口味偏好、人数和烹饪时间, 从内置食材库和菜谱库中智能匹配,输出完整的菜单方案和分步烹饪指南。 当用户提到"今天吃什么""做什么菜""菜单""菜谱推荐""晚饭吃啥"等意图时触发。 It is an AI Agent Skill for Claude Code / OpenClaw, with 217 downloads so far.
How do I install Lesson?
Run "/install lesson" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Lesson free?
Yes, Lesson is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Lesson support?
Lesson is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Lesson?
It is built and maintained by yequanzheng (@yequanzheng); the current version is v0.1.1.
More Skills