← 返回 Skills 市场
enawareness

Run Coach

作者 LeoYann · GitHub ↗ · v1.0.7 · MIT-0
cross-platform ⚠ suspicious
286
总下载
0
收藏
0
当前安装
8
版本数
在 OpenClaw 中安装
/install run-coach
功能描述
Science-based running coach with HD visual training plans and Garmin sync. For all runners — from 5K fitness to marathon.
安全使用建议
What to consider before installing: - The skill requires a Telegram bot token (TELEGRAM_BOT_TOKEN). Anyone with that token can control your bot, so create a dedicated bot and keep the token private. Use a dedicated chat id or account for the bot to minimize exposure. - Garmin integration is optional but requires GARMIN_EMAIL and GARMIN_PASSWORD; if you enable it the unofficial garminconnect library will be used and a session token file (garmin/.garth) will be stored on disk. Only provide Garmin credentials if you understand and accept this risk (and consider revoking passwords or tokens afterwards). - The skill runs local scripts (bash + Node + Python). That means code included in the skill will execute on the host when you ask the agent to render/send images or sync Garmin data — run it in an isolated container or environment you control (not on a shared host) if you have concerns. - The SKILL.md tells you to install system packages (fonts, Playwright Chromium) if you want the HD image pipeline. Those are user-run steps — validate them and prefer isolated environments. - If you want extra caution: inspect the scripts locally, create a dedicated Telegram bot and user account for this skill, avoid storing highly privileged or reuseable secrets in global envs, and only run garmin-sync when needed. If anything about the skill's behavior is unexpected, stop and inspect the garmin/ and training/ directories and the contents of the files written there.
功能分析
Type: OpenClaw Skill Name: run-coach Version: 1.0.7 The 'run-coach' skill is a legitimate tool for tracking running activities and generating visual training plans. It uses the official Telegram Bot API and the 'garminconnect' library to sync user data, with all sensitive credentials (tokens and passwords) managed via environment variables as per standard OpenClaw practices. The code logic in 'training/text-to-image.sh' and 'training/screenshot.mjs' is transparent, focusing on rendering HTML to images using Playwright, and lacks any indicators of data exfiltration, persistence, or malicious intent.
能力评估
Purpose & Capability
Name/description (Telegram coach + optional Garmin sync + image pipeline) align with the included files and envs: Telegram bot token + chat id are required and node is used for the image/sending pipeline. Garmin sync code exists and is described as optional in SKILL.md, which matches that GARMIN_EMAIL/ GARMIN_PASSWORD are not required by the skill metadata but are documented for optional use.
Instruction Scope
SKILL.md directs the agent to execute local scripts (text-to-image.sh, send-plan.sh and the Node/Python helpers) and to read/modify MEMORY.md for the user profile. That scope matches the stated features. Note: the garmin sync script will read GARMIN_EMAIL/GARMIN_PASSWORD from env and will perform network calls to Garmin and write token/cache and JSON activity files to the garmin/ directory; these actions are limited to the stated optional Garmin functionality but will persist credentials/tokens on disk if used.
Install Mechanism
There is no automatic install spec (instruction-only), which reduces supply-chain risk. SKILL.md instructs you to run apt-get, pip install, and npx playwright install if you want images — these are user-run instructions rather than automatic downloads from unknown hosts. Playwright and garminconnect are public packages; the files included are local scripts that call those tools.
Credentials
Required env vars declared in metadata are just TELEGRAM_BOT_TOKEN and TELEGRAM_CHAT_ID — appropriate for Telegram-based operation. Garmin credentials are documented as optional in SKILL.md (GARMIN_EMAIL, GARMIN_PASSWORD) and are only needed for the optional garmin-sync script; however, those optional envs are not listed in the metadata requires.env. GARMIN_PASSWORD is sensitive and, if provided, will be used by an unofficial garminconnect library and a session token will be cached to disk in garmin/.garth.
Persistence & Privilege
always is false and the skill does not request system-wide persistence. The garmin sync implementation caches session tokens and writes JSON activity files under the skill's garmin/ directory — normal for a sync tool but worth noting for privacy. The skill does not modify other skills or global agent settings.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install run-coach
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /run-coach 触发
  4. 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
v1.0.7
- Removed Garmin environment variables (GARMIN_EMAIL, GARMIN_PASSWORD) from required list; Garmin sync is now fully optional. - Updated metadata to reflect only required environment variables (Telegram). - No changes to user-facing features or instructions.
v1.0.6
- Added optional GARMIN_EMAIL and GARMIN_PASSWORD fields to metadata for easier Garmin Connect sync. - Clarified that image-based run logging uses LLM vision input; no OCR scripts are included or needed. - Minor instructions updates for environment setup and image pipeline. - Improved documentation to clarify which visual/vision features are native and which require external scripts.
v1.0.5
- Streamlined setup by removing training/setup.sh and updating install instructions for dependencies. - Apt package installation and font setup are now manual steps, not automated. - Metadata updated: removed package requirements and consolidated system dependencies guidance. - SKILL.md cleaned up and clarified; documentation is more concise and easier to follow. - Core functionality and usage are unchanged.
v1.0.4
Version 1.0.4 - Simplified Garmin integration setup: `garminconnect` now declared as an optional pip dependency in the manifest. - Updated SKILL.md environment variable list to reflect that only `TELEGRAM_BOT_TOKEN` and `TELEGRAM_CHAT_ID` are required; Garmin credentials are optional. - Environment and package installation instructions are clearer and easier to follow.
v1.0.3
- Improved robustness of the training/text-to-image.sh script for converting text to image. - Minor adjustments to training/setup.sh to streamline dependency installation and setup process.
v1.0.2
- Weekly training plans are now sent as a 2-photo album (run days + cross-training days) for improved organization. - Image album sending pipeline updated: training/send-album.mjs and send-plan.sh now default to Telegram sendPhoto/sendMediaGroup, removing ImageMagick split. - Minor SKILL.md simplifications: removed unneeded "convert", "curl", and "identify" dependencies; updated instructions for plan/image sending. - Agent instructions clarified for sending multi-photo and single-photo plans.
v1.0.1
- Improved reliability and stability of the image rendering and album sending pipeline. - Enhanced error handling in training/screenshot.mjs and training/send-album.mjs for better user feedback when issues occur.
v1.0.0
Initial release of run-coach: a science-based running coach for Telegram with Garmin sync and HD visual training plans. - Log runs via text or by sending screenshots from any running app/device — no integration required. - Visual training plans are rendered as HD photo albums, optimized for Telegram Photos tab (no compression, emoji and CJK support). - Automatically sync and analyze Garmin data (optional), including trend tracking and injury monitor. - Generate weekly summaries, 4-week progress reviews, and personalized race prep for all distances. - Detailed setup and system dependency instructions provided for smooth installation. - Built-in evidence-based training methodologies: Daniels VDOT, MAF HR, FIRST, and 80/20 structure.
元数据
Slug run-coach
版本 1.0.7
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 8
常见问题

Run Coach 是什么?

Science-based running coach with HD visual training plans and Garmin sync. For all runners — from 5K fitness to marathon. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 286 次。

如何安装 Run Coach?

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

Run Coach 是免费的吗?

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

Run Coach 支持哪些平台?

Run Coach 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 Run Coach?

由 LeoYann(@enawareness)开发并维护,当前版本 v1.0.7。

💬 留言讨论