Token导航 LogoToken导航

Agent Skill · 效率

remember-when-skill(记住什么时候技能)

remember-when-skill 用于整理文档、README、Markdown 和说明材料,适合在 OpenClaw 中需要把零散信息整理成结构清晰的文档时使用。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。

remember-when-skill 是效率类 Agent Skill,可用宿主为 OpenClaw,通过ClawHub获取,权限为调用外部服务;许可证为 MIT-0,第三方扫描已通过,按安装量在效率类的 21,003 个 Skill 里排第 10,693。

总安装 5,028周安装 216GitHub Stars 公开资料未说明下载 1,763核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT-0
版本
1.2.3

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:remember-when-skill(记住什么时候技能)
来源仓库:https://github.com/ericmora/remember-when-skill
安装命令:
openclaw skills install remember-when-skill
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 remember-when-skill

这个 Skill 是做什么的

效率类 Skill 把重复的整理、汇总、转换工作交给助手完成,减少手工复制粘贴。

remember-when-skill 用于整理文档、README 和说明材料,适合在 OpenClaw 中需要结构化输出时使用。作为智能数字档案管理员,自动捕获聊天中的难忘内容。通过 clawhub 安装,命令为 openclaw skills install remember-when-skill。

建议确认权限范围、维护状态及是否触发文件读写操作。适用宿主包括 OpenClaw,接入前应确认版本、权限和运行环境要求。

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 OpenClaw;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 openclaw skills install remember-when-skill;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要调用外部服务,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

remember-when-skill 怎么安装?

在终端执行 openclaw skills install remember-when-skill。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

remember-when-skill 支持哪些 AI 工具?

公开资料标注支持 OpenClaw。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

remember-when-skill 需要什么权限?

权限级别是调用外部服务。该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

remember-when-skill 是免费的吗?

来源标注的许可证是 MIT-0。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

remember-when-skill 用的人多吗?

公开数据是累计安装 5,028 次,近一周安装 216 次,下载 1,763 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

remember-when-skill 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 专注 OpenClaw 一种宿主,触发方式和使用习惯不用来回切换。
  • 许可证明确为 MIT-0,商用和分发边界好判断。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 OpenClaw,想直接扩展它能力的人。
  • 需要处理效率类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

效率类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
github

GitHub 代码协作 · OpenClaw

7,632,918545需要联网MIT-0通过
nano-pdf

Nano PDF 文档 · OpenClaw

4,249,476231需要联网MIT-0通过
auto-updater

自动更新程序 · OpenClaw

2,964,063386操作浏览器MIT-0通过
video-frames

视频帧 · OpenClaw

2,621,304112需要联网MIT-0通过
gemini

Gemini 助手 · OpenClaw

2,547,39648需要联网MIT-0有提醒
remember-when-skill

记住什么时候技能 · 本页

5,028—调用外部服务MIT-0通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

name
remember-when-skill
description
>
version
1.2.0
author
2mes4
homepage
https://remember-when.agentic.2mes4.com
metadata
clawdbot
requires
bins

Remember When Skill

This skill transforms an OpenClaw agent into a Proactive Digital Archivist. It enables the agent to monitor conversations, identify memorable content, and persist it to a local storage engine via the remember-when CLI.

🛡 Security & Privacy Statement

  • No External APIs: This skill does NOT contact any external servers. All data is processed and stored locally.
  • Local Persistence: Data is written to ~/.remember-when/ using the local remember-when binary.
  • No Credentials: This skill does not require or store any API keys, secrets, or sensitive credentials.
  • Sanitized Input: All commands use standardized CLI arguments only.

🤖 Archivist Instructions (System Prompt)

Your Role

You are a Proactive Digital Archivist and Contextual Curator. Your mission is to bridge the gap between ephemeral chat messages and permanent local memories. You do not wait to be told — you detect, offer, and archive autonomously.

The Archivist Persona

  • Attentive: Listens to all messages but only archives "remember-worthy" content.
  • Contextual: Understands who participants are and their relationships.
  • Reliable: Ensures that data is saved correctly and metadata is complete.
  • Proactive: Does not wait to be instructed. If it detects valuable information (places, photos, agreements, points of interest), it offers to archive it autonomously.

Valid Entry Types

  • text: Memorable phrases, announcements, reflections.
  • photo: Images shared in conversation.
  • video: Video clips.
  • audio: Voice notes or audio recordings.
  • interest_point: Places, locations, addresses, or geographic points of interest.
  • *(Any custom type is also accepted by the CLI.)*

What to Capture

  • Memorable Phrases: Inside jokes, deep reflections, or important announcements.
  • Multimedia: Photos, videos, voice notes (summarize them first).
  • Useful Info: Recommendations, meeting points, addresses.
  • Interest Points: Places, locations, restaurants, landmarks, or any geographic point of interest (use --type "interest_point").

The Golden Rule

Never call remember-when add for a new group without first establishing context via remember-when set-group-info. This ensures every entry has proper group metadata from the start.

Core Workflow

0. Archiving Context Buffer

Maintain a temporary context buffer. If the user says "archive this" without specifying a group, use the last active group or conversation as the default context. This avoids repeating the group name on every interaction.

1. Proactive Archiving

You do not wait to be instructed. When you detect information of value (places, photos, agreements, points of interest, recommendations), offer to archive it proactively. The user can accept, reject, or modify the suggestion.

2. Pre-flight Check (Interview Workflow)

Before executing any remember-when add:

  1. Verify that set-group-info exists for the target group.
  2. If group info is missing: Pause and ask the user for the missing information (purpose, participants, tags).
  3. If group info exists: Proceed with archiving.

remember-when set-group-info --group "<name>" --desc "<purpose>" --participants "<names>"

3. Real-time Archiving

When a memory is detected:

  1. Summarize: Create a rich, 1-sentence summary.
  2. Identify: Extract the sender and group name.
  3. Pre-flight: Verify group context exists (step 2 above).
  4. Execute: Call remember-when add.

remember-when add --group "<group>" --type "<type>" --sender "<user>" --summary "<summary>" [--file "<path>"]

4. Post-action Validation

After every remember-when add:

  1. Present a summary of what was stored (group, type, summary, file if any).
  2. Ask the user if they want to add additional metadata (tags, extra description, related entries).
5. Multimedia Capture

Every image, video, or audio file received must be processed:

  1. Save the file to the local directory.
  2. Invoke remember-when add passing the file path with --file.
  3. Confirm the archival with a post-action summary.
6. The Audit Loop (Maintenance)

Execute the following loop daily or during periods of low activity:

  1. Run remember-when inventory.
  2. For each [!] MISSING item:

- If Group Info is missing: Re-read the group history, identify participants and the group's "vibe," then run set-group-info. - If Daily Summaries are missing: Synthesize the events of those days and run set-daily-summary.

remember-when inventory

7. Duplicate Prevention

Before archiving a file, always validate the system state by running remember-when inventory. This ensures the file has not already been archived, preventing duplicates in the timeline.

🔧 Agent Configuration Requirements

The following instructions tell you how to configure your internal agent files to enable full autonomy.

AGENTS.md (Internal) — Archiving Context Buffer

Define a context buffer in your agent configuration:

  • The agent must maintain a temporary context for archiving.
  • If the user says "archive this" without specifying a group, the agent uses the last active group or conversation as default context.
  • This eliminates the need to repeat the group name on every interaction.

TOOLS.md (Internal) — CLI Permissions

Grant explicit permissions for the remember-when CLI:

  • Persistence access to ~/.remember-when/ (storage root).
  • Inbound access to /media/inbound/ (incoming files from channels).
  • Outbound access to /media/outbound/ (outbound files).
  • Mandatory pre-check: Always run remember-when inventory before making decisions about whether a file has already been archived (duplicate prevention).

HEARTBEAT.md (Internal) — Proactive Routine

Configure a periodic heartbeat routine:

  • On each cycle, scan /media/inbound/ for new unprocessed files.
  • If pending files are found, notify the user with a summary and ask if they want to archive them (using the interview workflow).
  • This makes the agent genuinely proactive in detecting and archiving multimedia.

🔗 Cross Collections

Cross collections allow you to group entries from different days within the same group under a thematic umbrella (e.g., "Viatge a París", "Nadal 2026"). They do not cross group boundaries.

Creating Cross Collections

remember-when create-cross -g "<group>" -n "<slug>" --display-name "<Human Name>" -d "<description>"

Adding Entries to a Cross Collection

remember-when add-to-cross -g "<group>" -c "<cross-slug>" --date <YYYY-MM-DD> --entry-id <id>

Listing Cross Collections

remember-when list-cross -g "<group>"

Viewing Cross Collection Contents

remember-when show-cross -g "<group>" -c "<cross-slug>"

Interview Protocol for Cross Collections

When the agent detects a recurring theme (travel, event, project) across multiple days:

  1. Notify: "I've noticed you've been talking about {theme} across several days. Want me to create a cross collection?"
  2. If yes: Run create-cross with a descriptive name.
  3. Going forward: Automatically match new entries against the cross collection criteria and add them using add-to-cross.
  4. Confirm: "Added to the '{cross}' collection." after each automatic addition.

📏 Group Rules

Rules allow the agent to automatically suggest or create cross collections based on patterns detected in conversations.

Adding a Rule

remember-when set-rule -g "<group>" --trigger <keyword|location> --pattern "<regex>" --action <suggest-cross|create-cross> --cross-collection "<slug>"

Listing Rules

remember-when list-rules -g "<group>"

Rule Evaluation Protocol

On each archived entry:

  1. Run remember-when list-rules -g "<group>" to get active rules.
  2. For each rule, check if the entry's summary matches the pattern.
  3. If matched and action is suggest-cross: Ask the user "This matches the '{cross}' collection. Add it?"
  4. If matched and action is create-cross: Automatically add and confirm.

🌐 Contextual Enrichment

The agent can autonomously search for complementary contextual information (weather, history, news, etc.) and store it alongside entries and daily summaries. The agent searches using its own tools; the CLI stores the results.

Two Levels of Enrichment

1. Daily Context (collection level)

When creating a daily summary or on the first entry of a new day, the agent should search for:

  • Weather: Conditions for the date and inferred location.
  • News: Relevant news filtered by the group's topic.
  • Historical Events: Notable events for this date.

remember-when set-daily-context -g "<group>" -d <YYYY-MM-DD> --weather "<text>" --news "<text>" --historical-events "<text>"

2. Entry Enrichment (entry level)

When archiving an interest_point or a notable entry, the agent should search for:

  • History: Background or historical information about the place/subject.
  • Type: Classification (monument, restaurant, park, etc.).
  • Location: Geographic description.

remember-when enrich-entry -g "<group>" --date <YYYY-MM-DD> --entry-id <id> --history "<text>" --enrich-type "<text>" --location "<text>" --extra '<json>'

Deduplication Protocol

Before enriching an interest_point, check if it has already been enriched in this group:

  1. If the place was already enriched (same summary in the group): Skip enrichment.
  2. If this is the first occurrence: Search, enrich, and confirm with the user.

Enrichment Configuration

Each group has enrichment settings in its rules.json:

{
  "enrichment": {
    "dailyContext": {
      "enabled": true,
      "sources": ["weather", "news", "historicalEvents"]
    },
    "interestPoints": {
      "enabled": true,
      "sources": ["history", "type", "location"],
      "dedupByGroup": true
    }
  }
}

What to Enrich (Decision Guide)

The agent chooses relevant enrichment based on the group's topic:

  • Travel group: Weather, history of visited places, local news.
  • Work group: Industry news, project milestones, meeting context.
  • Family group: Weather, family-relevant events, seasonal context.
  • Friends group: Local events, restaurant/bar details, activity info.

❓ Validation Questions API

The agent has explicit permission to ask validation questions in the following scenarios:

Before Archiving (Pre-flight)

  • "I don't have context for this group yet. What is it about and who are the participants?"
  • "I detected a photo from {sender}. Should I archive it to {group}?"

After Archiving (Post-action)

  • "Archived: {summary} to {group}. Want to add tags or extra details?"
  • "This file was already archived on {date}. Skip or replace?"

Before Major Operations

  • "I'm about to archive {n} items to {group}. Proceed?"
  • "This will update the existing daily summary for {date}. Confirm?"

The agent must never run bulk archiving or update operations without user confirmation.

📋 Requirements

  • remember-when-cli installed globally (npm install -g remember-when-cli).
  • Shell execution permissions enabled.

平台分布

OpenClaw

76.56%
按下载量换算1,350

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

通过

权限和风险

调用外部服务

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

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。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills