---

name: fitness-skill description: "运动健康管家 Pro — 专为 Open Claw + 飞书生态打造的 AI 健身管家。双模式打卡（事后总结 / 边练边记）、3 层 Session 防丢失、飞书云文档同步、极致 Token 优化。Route to this skill when the user mentions workouts, logs exercises, asks for fitness stats, or provides post-workout feedback." version: 1.2.2 metadata: {"clawdbot":{"emoji":"💪","requires":{"bins":["python3"],"env":[]},"os":["linux","darwin"]}}

运动健康管家 Pro (Fitness & Workout Manager Pro)

⚠️ 平台要求：本 Skill 依赖「飞书 (Feishu)」生态，需为 Open Claw 绑定的机器人开通消息读写与云文档管理权限。

专为 Open Claw 打造的 AI 健身管家。无论是健身房力量训练还是网球对战，通过自然语言即可完美记录——你最省心的赛博教练。

✨ 核心亮点

• 双模式无缝打卡 — 事后一句话总结（"打网球 2 小时感觉很累"）或沉浸式边练边记（"卧推 60kg 8个 3组"），AI 自动解析入库。
• 防遗忘 Session 保护 — 3 层生命周期管理（主动关怀 → 心跳巡检 → 硬截止），练完忘了说"结束"也能自动闭环，绝不丢失数据。
• 飞书全景同步 — 计划 & 周/月度统计自动推送至飞书云文档，内置 Cron 定时提醒。
• 极致 Token 优化 — 日常查询仅返回紧凑摘要，全量记录自动导出为 Markdown，拒绝上下文爆炸。

🤖 Agent 路由指南

• Capabilities: Stateful session handling (fitness-log start/end), NLP workout parsing, localized JSON storage, Feishu document sync, auto-stale-session cleanup.
• Trigger: Route here when the user mentions workout routines, logs exercises, asks for fitness stats (fitness-status), or provides unstructured post-workout feedback.

Quick Start

# Create a fitness plan
fitness-plan create --goals "muscle_gain,cardio" --sport_pref "gym,tennis" --frequency 4

# One-shot log (after workout)
fitness-log "gym: bench press 60kg x 8 x 3, squat 80kg x 5 x 4, 1.5 hours"

# Live session (during workout — log each exercise as you go)
fitness-log start gym
fitness-log "bench press 60kg x 8 x 3"
fitness-log "squat 80kg x 5 x 4"
fitness-log "felt great"
fitness-log end "good session overall"

# View status
fitness-status --week
fitness-status --month
fitness-status --export

Commands

fitness-plan

Command	Description
fitness-plan init	Interactive multi-turn Q&A to generate a plan
fitness-plan create --goals ... --frequency N	Non-interactive plan creation
fitness-plan show	Display the current plan (Markdown)
fitness-plan update --frequency 4 --level advanced	Modify an existing plan
fitness-plan sync	Push plan to Feishu doc

fitness-log

Two modes: one-shot (after workout) and live session (during workout).

Command	Description
fitness-log "<text>"	One-shot: parse and log a single workout entry
fitness-log start [type]	Start a live session (e.g. start gym)
fitness-log "<exercise text>"	Append exercise to active session
fitness-log end ["notes"]	End session, auto-calculate duration, commit to log
fitness-log status	Show active session info (includes idle/stale warnings)

Live session workflow: The user sends messages during training. Each message is appended to the same session entry. When the user says "end", all exercises are merged into one log entry with auto-calculated duration (from start to end time).

fitness-status

Command	Description
fitness-status	Last 7 entries (compact, token-efficient)
fitness-status --week	Weekly summary with stats
fitness-status --month	Monthly summary (last 30 days)
fitness-status --days N	Custom period summary
fitness-status --all	All entries (capped at 50, with truncation notice)
fitness-status --export [N]	Export last N days (default 30) as Markdown file

Session Lifecycle & Auto-Close

Sessions that are not explicitly ended are handled automatically:

Condition	Behavior
Idle ≥ 30 min	Claw proactively asks the user if training is done (via AGENTS.md rule). At most once per conversation to avoid nagging.
Idle > 4 hours or next day	Session is marked stale and auto-closed on the next fitness-related action. Duration is calculated up to the last message (not current time).
Heartbeat detects stale session	HEARTBEAT.md rule runs fitness-log end to force-close, and Claw notifies user on next interaction.
New start while stale exists	Old stale session is auto-closed first, then new one begins.

This ensures no workout data is ever lost, even if the user forgets to say "end".

Token Efficiency

• Default queries return compact summaries (7 entries or period stats) to minimize token usage.
• --all is capped at 50 entries with a truncation notice.
• For full history, use --export to generate a Markdown file that can be sent to the user directly, avoiding large inline output.

Data Storage

• Plan: ~/.openclaw/workspace/fitness-skill/plan.json
• Log: ~/.openclaw/workspace/fitness-skill/log.json
• Active session: ~/.openclaw/workspace/fitness-skill/active_session.json (auto-deleted on end)
• Exports: ~/.openclaw/workspace/fitness-skill/exports/

Feishu Integration

• Set FEISHU_FITNESS_DOC_ID to enable plan/log sync to a Feishu doc.
• sync_plan_to_feishu() — push plan Markdown
• sync_log_to_feishu(days=30) — push log report Markdown
• Daily reminders via scheduler.add_cron_job() at the user's configured time.