功能描述

专属健身追踪与饮食记录助手。支持体重/饮食/运动/饮水记录、BMR/TDEE自动计算、碳循环等饮食策略、食物数据库自动计算营养素、营养知识问答、训练计划制定、趋势图表、定时提醒。触发词：记录饮食/体重/运动/饮水、健身打卡/设置/初始化、今天吃了/练了、查看进度、训练建议、喝水了吗、蛋白质/碳水/脂肪/热量相关、...

使用说明 (SKILL.md)

fitbuddy — Fitness Buddy

Name: FitBuddy
Author: promisewhh

⛔ 绝对规则

数据路径：skills/fitbuddy/fitbuddy-data/。脚本 cwd 为 skills/fitbuddy/
先读 profile 再行动：profile.json 有 BMR/TDEE/宏量/饮食策略，不要问用户已有的信息
输出规范：只输出最终用户消息，禁止暴露思考过程/步骤/内部逻辑
安全红线：热量缺口 ≤ 20% TDEE，减脂速度 ≤ 1kg/周，三大营养素缺一不可

初始化

如果 fitbuddy-data/profile.json 不存在 → 读 references/init-guide.md

用户意图 → 执行

用户说	执行
记录体重	`record.py weight --date 今天 --kg 数值` → 更新 profile → 汇报+对比昨天
吃了什么	`record.py meal --date 今天 --food-name "食物" --grams 克数` → 汇报热量/宏量/剩余
吃了菜品（番茄炒蛋等）	`record.py meal --date 今天 --food-name "番茄炒蛋" --grams 200` → 菜品自动拆解食材
模糊份量（一碗饭/一个鸡蛋）	AI 侧翻译：一碗饭→`--food-name "白米饭" --grams 200`，一个鸡蛋→`--grams 50`，一个鸡腿→`--grams 120`，一根香蕉→`--grams 100`
吃了多个食物	逐条调用 `record.py meal`，每条自动保存
跟昨天一样	`record.py meal --date 今天 --like-yesterday` → 追加昨日餐食
记错了/撤销	`record.py undo --date 今天`
还剩多少/今天吃了啥	`record.py today --date 今天` → 读取后汇报
看进度	`record.py progress --date 今天` → 进度条
看周报	`record.py weekly --date 今天`
看常吃什么	`record.py patterns`
喝水	`record.py water --date 今天 --ml 数值`
运动了	`record.py exercise --date 今天 --name 名称 --sets 组 --reps 次 --weight 重量 --group 肌群 --type strength`
改设置/改目标	直接改 profile.json
训练建议	读 references/exercise.md + references/training-plan.md
营养知识	读 references/nutrition-guide.md
该吃啥/推荐	读今日记录 + `calc.py daily` 算剩余 → 推荐补蛋白质缺口的食物组合

关键机制

餐次自动判断：--meal 不填时按时间自动归类（早6-10/午10-14/加14-16/晚16-21）

默认份量：food-db.json 的 serving_g 字段。--grams 不填时自动使用

食物匹配：food-db search 模糊匹配 → 找到用DB数据 → 找不到则估算营养素并用 food-db add 存入DB

菜品：food-db 中 type: dish 的条目，记录时自动按比例拆解成食材计算

碳日目标：python scripts/calc.py daily --profile fitbuddy-data/profile.json --date 日期 返回当天热量/宏量目标

定时提醒

逻辑统一在 references/reminder-logic.md。

Cron 任务：执行 fitbuddy 定时提醒。读取 skills/fitbuddy/references/reminder-logic.md，按照 \x3CTYPE> 类型的逻辑执行。

类型：weight / breakfast / lunch / dinner / water / preworkout

数据文件

fitbuddy-data/
  profile.json              — 用户档案
  records/YYYY-MM-DD.json   — 每日记录
  food-db.json              — 食物数据库
  user-patterns.json         — 用户习惯（自动维护）

参考文档

文件	内容
references/init-guide.md	初始化引导
references/nutrition-guide.md	营养知识&份量估算
references/nutrition.md	公式参考
references/exercise.md	运动指南
references/training-plan.md	训练计划
references/budget-meals.md	平价饮食方案
references/reports.md	报告&图表
references/channels.md	提醒频道配置
references/reminder-logic.md	提醒逻辑

安全使用建议

This skill appears coherent with its stated purpose. Before installing: (1) Review the bundled scripts (especially record.py) to verify you are comfortable with local file writes under skills/fitbuddy/fitbuddy-data/ (profile.json and records/). (2) Be aware the skill may create OpenClaw cron tasks to send reminders; creating/removing cron tasks can require gateway/device permissions — approve only if you trust the skill and understand pairing implications. (3) Optional McDonald's/DingTalk integrations require you to add tokens/config (mcporter or mcporter.json); these are opt-in. If you enable them, store tokens securely and confirm promo/auto-claim features remain off (they default to OFF). (4) If you want minimal surface, do not enable external channel integrations and keep reminders disabled or manually configured. If you want higher assurance, run the skill in a sandboxed agent instance and inspect the created cron tasks and config files after initialization.

功能分析

Type: OpenClaw Skill Name: fitbuddy Version: 2.1.1 The fitbuddy skill bundle is a comprehensive fitness and nutrition tracking tool designed for the OpenClaw platform. It includes Python scripts (record.py, calc.py, chart.py, init_profile.py) for managing user data, calculating health metrics, and generating progress visualizations using matplotlib. The SKILL.md file provides clear operational instructions and safety guardrails, such as limiting calorie deficits and weight loss rates. The bundle features a reminder system using OpenClaw's cron functionality and supports optional integrations with external services like DingTalk and McDonald's for nutrition data, all of which are documented with user-controlled configuration in mind. No evidence of malicious intent, data exfiltration, or harmful prompt injection was found.

能力评估

✓ Purpose & Capability

Name/description (fitness tracking, logging, reminders, nutrition) matches the provided scripts, data files, and reference docs. Optional McDonald's MCP integration is documented and justified by the feature; no unrelated credentials or services are required by default.

✓ Instruction Scope

SKILL.md instructs the agent to read and update local files under skills/fitbuddy/fitbuddy-data/, run included Python scripts (calc.py, record.py, chart.py, init_profile.py) and create cron reminders. Those operations are within the stated purpose; the skill does not instruct broad system inspection or exfiltration of unrelated data. The only external interactions are optional channel integrations (mcporter/MCP, DingTalk/WeChat) described for sending reminders.

✓ Install Mechanism

No install spec is provided (instruction-only with bundled scripts). That is low-risk; included Python scripts run locally and require Python/matplotlib as documented. There are no downloads from arbitrary URLs or package installs embedded in the skill packaging itself.

✓ Credentials

The skill declares no required environment variables or primary credential. Optional integrations (McDonald's MCP, DingTalk MCP) require user-supplied tokens/configuration and are explicitly documented as opt-in. No secret is required by default, and the documentation explains where tokens would be stored if the user enables integration.

ℹ Persistence & Privilege

The skill creates cron reminders (OpenClaw cron) and writes/reads data under its own skills/fitbuddy/fitbuddy-data/ directory (profile.json, records, food-db). This is consistent with reminder and tracking functionality but does mean the skill will persist data locally and may create scheduled tasks — users should review and approve cron creation and any gateway/device pairing steps.

版本历史

v2.1.1

SKILL.md精简：删除重复内容、噪音文件引用、未实现功能；模糊份量改为AI侧翻译

v2.1.0

v2.1: 菜品级记录、用户习惯学习、模糊份量、进度可视化、周报、时间戳undo、like-yesterday追加模式

v2.0.0

v2.0大改造：批量记录、自动餐次、撤销、默认份量、SKILL.md重写、cron统一逻辑、食物库扩充22种、体重趋势分析

v1.2.0

Added McDonald's MCP integration (optional), i18n support, smart reminders, food database, English SKILL.md

v1.1.0

i18n support, smart reminders, food database, English SKILL.md

v1.0.0

Initial release

元数据

Slug fitbuddy

版本 2.1.1

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 6

常见问题

FitBuddy 是什么？

专属健身追踪与饮食记录助手。支持体重/饮食/运动/饮水记录、BMR/TDEE自动计算、碳循环等饮食策略、食物数据库自动计算营养素、营养知识问答、训练计划制定、趋势图表、定时提醒。触发词：记录饮食/体重/运动/饮水、健身打卡/设置/初始化、今天吃了/练了、查看进度、训练建议、喝水了吗、蛋白质/碳水/脂肪/热量相关、... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 131 次。

如何安装 FitBuddy？

在 OpenClaw 或 Claude Code 对话框中运行命令「/install fitbuddy」即可一键安装，无需额外配置。

FitBuddy 是免费的吗？

是的，FitBuddy 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。

FitBuddy 支持哪些平台？

FitBuddy 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 FitBuddy？

由 PromiseWHH（@promisewhh）开发并维护，当前版本 v2.1.1。

FitBuddy