Token导航 LogoToken导航TokenDH.com
效率敏感数据clawhub未标认证来源可访问clear审计通过

medication-channel用药频道

Agent Skill

medication-channel 用于补充效率相关能力,适合在 OpenClaw 中需要让 Agent 承接效率相关任务时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

2,661

周安装

112

GitHub Stars

公开资料未说明

下载量

932
OpenClaw

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

MIT-0

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:medication-channel(用药频道)
来源仓库:https://github.com/cdmichaelb/medication-channel
安装命令:
openclaw skills install medication-channel
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install medication-channel

简介

medication-channel 通过确定性路由机制处理药物事件,提升用药跟踪的准确性。

  • 适合需要自动化管理处方流转、药品提醒和用药记录的临床或家庭场景。
  • 输入药物名称、剂量或使用时间,系统按预设规则分配处理通道并返回执行状态。
  • 使用前需确认权限范围和集成方式,建议查阅原始仓库了解配置细节。
  • 该技能不涉及药物推荐,仅辅助流程管理,重要用药请咨询专业医师。

SKILL.md

name
medication-channel
version
1.0.0
description
Handle messages in a medication-tracking channel by routing medication events through a deterministic medication tracker instead of freehand model judgment. Use when working in a dedicated medication channel, especially for logging taken/missed/extra medication events, morning/evening completion, timestamp-grounded confirmations, and avoiding hallucinated reminders or fake times.
env
MEDICATION_TIMEZONE
description
IANA timezone for local display (e.g. America/Los_Angeles)
required
true
WORKSPACE
description
Root workspace directory for data files
required
false

Medication Channel

Use the deterministic medication logging path whenever source message metadata is available.

Configuration

Set the following environment variables or replace placeholders before use:

  • MEDICATION_TIMEZONE — IANA timezone string for local time display (e.g. America/Los_Angeles, Europe/Berlin)
  • WORKSPACE — root directory for data files (defaults to ~/.openclaw/workspace)

Medication names and schedules are configured at the top of scripts/tracker_v2.py — edit the MORNING_MEDS, EVENING_MEDS, and KNOWN_MEDS lists to match your regimen.

Core Rules

  • Treat the configured channel as the dedicated medication channel.
  • Prefer the log_from_discord.sh wrapper or tracker_v2.py log-from-message.
  • Never use now as a logged/displayed time.
  • Never derive event time from the model turn time.
  • Use the source Discord message timestamp unless the user explicitly gives a different time.
  • Convert displayed times to the configured timezone.
  • Never log reminder chatter as medication history.
  • Never treat an assistant reminder message, assistant follow-up, assistant summary, or assistant self-referential chatter as a medication event.
  • Never emit a confirmation like "taken", "logged", or "recorded" in response to an assistant-generated reminder unless a real user medication report was just logged through the script path.
  • Never claim something was logged unless the script path actually ran.
  • Prefer raw-message passthrough to the tracker instead of model-side parsing.

Preferred Workflow

Log a medication message from Discord

Use when the user reports:

  • taken / took
  • done / completed
  • missed / skipped
  • extra dose
  • clear natural-language medication intake

Preferred path:

scripts/log_from_discord.sh \
  "$MESSAGE_TEXT" \
  "$CHANNEL_ID" \
  "$MESSAGE_ID" \
  "$AUTHOR_ID" \
  "$TIMESTAMP_UTC"

Equivalent direct call:

python3 scripts/tracker_v2.py log-from-message \
  "$MESSAGE_TEXT" \
  "$CHANNEL_ID" \
  "$MESSAGE_ID" \
  "$AUTHOR_ID" \
  "$TIMESTAMP_UTC"

Then use the returned row or format-confirmation output for the confirmation.

Parse only

Use only when inspecting behavior or debugging.

python3 scripts/tracker_v2.py parse "$MESSAGE_TEXT" "$TIMESTAMP_UTC"

Format a confirmation

Use when you already have the JSON row and want the standard confirmation string.

python3 scripts/tracker_v2.py format-confirmation '<json-row>'

Special Guardrail: Reminder Messages

If the message being handled is itself a reminder, or the user is replying to/quoting a reminder message, do not assume medication was taken.

Examples that are not medication events by themselves:

  • an assistant message like Reminder: it's time to take your evening medication.
  • an assistant message like Evening medication taken (recorded at ...)
  • a user message like something went wrong again
  • a user reply that is discussing a reminder failure rather than reporting intake

In these cases:

  • do not call the tracker unless the user message itself contains an actual medication event
  • do not fabricate a confirmation
  • debug/explain briefly instead

What the Tracker Already Handles

The tracker is preferred because it already handles:

  • common natural-language medication phrases
  • normalization of common misspellings / aliases
  • morning/evening inference
  • dedupe by source Discord message id
  • local timezone display formatting
  • structured logging

So do not manually re-implement those steps unless debugging the tracker itself.

If Uncertain

Ask a short clarifying question only when needed, such as:

  • whether the message is a real medication event vs general discussion
  • whether the user is saying they skipped something or are uncertain
  • whether a custom explicit time should override the source timestamp

Data Access

This skill reads and writes persistent files in the workspace:

  • Writes $WORKSPACE/data/medication_log_v2.csv — one row per medication event (med name, status, timestamps, source message ID)
  • Writes deduplication state keyed by source message ID
  • Reads medication config from MORNING_MEDS/EVENING_MEDS/KNOWN_MEDS lists at top of scripts/tracker_v2.py
  • No network calls — fully local, stdlib-only Python
  • No external APIs — no vision, no remote services

Required Files

The following files are included in the bundle:

  • scripts/log_from_discord.sh — Discord wrapper script
  • scripts/tracker_v2.py — medication parser/logger
  • references/CHANNEL_RULES.md — optional channel behavior reference
  • references/README.md — optional setup reference

适合场景

01

OpenClaw 用户查找和安装 Skill 时

02

用户想查找某类 Agent Skill 时

03

需要根据任务场景推荐可安装能力包时

04

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

OpenClaw

73.85%
按下载量换算688

安全审计

VirusTotal

未展示

ClawScan

通过

Static analysis

通过

权限和风险

敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills