← 返回 Skills 市场
Quick Reminders
作者
Nikita Shkoda
· GitHub ↗
· v1.1.4
4919
总下载
9
收藏
18
当前安装
6
版本数
在 OpenClaw 中安装
/install quick-reminders
功能描述
Zero-LLM one-shot reminders (<48h) via nohup sleep + openclaw message send, operated via {baseDir}/scripts/nohup-reminder.sh.
安全使用建议
Install this only if you are comfortable with reminder text, timing metadata, and delivery targets being stored locally and sent through your configured OpenClaw messaging account. Avoid sensitive reminder contents, verify the channel and target in `TOOLS.md`, and use list/remove commands to review or cancel pending reminders.
功能分析
Type: OpenClaw Skill
Name: quick-reminders
Version: 1.1.4
The 'quick-reminders' skill is designed for one-shot reminders and appears benign. The `SKILL.md` provides clear instructions for the agent, including guardrails and message composition guidelines, without any evidence of prompt injection attempts to subvert the agent. The `nohup-reminder.sh` script uses `nohup` and `sleep` to schedule reminders, leveraging `jq` for JSON management and `openclaw message send` for delivery. It employs robust shell scripting practices, such as `set -euo pipefail`, file-based locking, atomic JSON updates, and careful argument handling for the `nohup bash -c` command to mitigate shell injection risks. Temporary message files are created with `umask 077` for security, and PIDs are validated before `kill` operations. The `README.md` provides transparent documentation on data access, security considerations, and limitations, confirming no credential handling or unauthorized data exfiltration. All operations align with the stated purpose of a reminder skill.
能力评估
Purpose & Capability
The skill, README, and script coherently implement one-shot reminders using local sleep processes and `openclaw message send`; multi-channel delivery is expected for the reminder purpose and not hidden.
Instruction Scope
The agent instructions restrict use to short-horizon reminders and direct 2+ day requests to a calendar, but the script itself does not enforce the 48-hour limit, so direct CLI use can exceed the advertised scope.
Install Mechanism
The artifact contains a SKILL.md, README, and local bash script with declared `jq` and `openclaw` dependencies; there is no remote installer, obfuscated setup, or package download behavior.
Credentials
The skill reads or writes `TOOLS.md`, `reminders.json`, and temporary message files to perform delivery; this is proportionate, but reminder content and contact identifiers may be stored in plaintext locally until cleanup.
Persistence & Privilege
Each reminder creates a detached `nohup` sleep process that later sends the message and removes its state; this persistence is central to the feature, documented, cancellable, and not reboot-persistent.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install quick-reminders - 安装完成后，直接呼叫该 Skill 的名称或使用
/quick-reminders触发 - 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
v1.1.4
- No user-visible changes;
- README updated.
v1.1.3
- Removed the _meta.json file from the skill.
- No changes to user experience or skill functionality.
v1.1.2
- Added _meta.json file for enhanced metadata support.
- Updated SKILL.md: "openclaw" is now a required binary dependency in metadata.
v1.1.1
Version 1.1.1
- Minor edit to Rule 4 for clearer instructions: "remove old → add new" changed to "remove old, then add new".
- No functional or command changes; behavior remains the same.
v1.1.0
Version 1.1.0
- Replaced unsafe bash -c string interpolation with argument-based detached runner execution (fixes injection/broken quoting risks for --channel, --target, paths)
- Added per-workspace temp message file naming using WORKSPACE_KEY (oclaw-rem-<workspace-key>-<id>.msg) to prevent cross-workspace collisions
- Added lock-based state protection (LOCK_DIR, acquire_lock, release_lock) around add, list, and remove to avoid concurrent JSON races
- Added atomic JSON writer (write_json with unique mktemp temp files) and replaced fixed .tmp writes
- Stored msg_file in each reminder record and updated removal logic to delete recorded file path reliably (with fallback cleanup for old filename format)
- Added pre-ID prune under lock so max(id)+1 runs on cleaned, synchronized state
- Hardened process/file handling:
- reminder message file now written with restricted perms via umask 077
- kill now guarded to numeric PIDs only
v1.0.0
- Initial release of quick-reminders: set human-friendly, one-shot reminders for up to 48 hours, delivered via openclaw CLI.
- Composes reminder messages at creation time (no LLM used at delivery).
- Strict <2 days guardrail — longer reminders are redirected to calendar with an explanation.
- Supports multiple channels (Telegram default; WhatsApp, Discord, Signal, iMessage via flag).
- CLI-only operation with easy commands for add, list, and remove.
- Reminder messages use a casual, friendly tone, never robotic.
- Confirmation responses are concise and future-tense (e.g., "Will do", "Got it").
元数据
常见问题
Quick Reminders 是什么？
Zero-LLM one-shot reminders (<48h) via nohup sleep + openclaw message send, operated via {baseDir}/scripts/nohup-reminder.sh. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 4919 次。
如何安装 Quick Reminders？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install quick-reminders」即可一键安装，无需额外配置。
Quick Reminders 是免费的吗？
是的，Quick Reminders 完全免费（开源免费），可自由下载、安装和使用。
Quick Reminders 支持哪些平台？
Quick Reminders 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 Quick Reminders？
由 Nikita Shkoda（@lstpsche）开发并维护，当前版本 v1.1.4。
推荐 Skills