Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

记一下

v0.2.0

像发消息一样记录一切(灵感/想法/知识/收支/日记/任务/引用),AI 自动分类、标签、关联,让知识自然生长。 支持微信/飞书消息输入,零摩擦记录。统一存储,多视图呈现(闪记视图/日记视图/周报视图)。 触发:用户发送任何想记录的内容时自动调用。

0· 97·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for yankj/just-note-ai.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "记一下" (yankj/just-note-ai) from ClawHub.
Skill page: https://clawhub.ai/yankj/just-note-ai
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install just-note-ai

ClawHub CLI

Package manager switcher

npx clawhub@latest install just-note-ai
Security Scan
VirusTotalVirusTotal
Pending
View report →
OpenClawOpenClaw
Suspicious
medium confidence
!
Purpose & Capability
The skill claims both a CLI ('just-note') and automatic WeChat/Feishu message ingestion via OpenClaw Gateway. However the package manifest refers to a bin/just-note executable (package.json and docs), but that executable is not present in the provided file list. Many docs/test reports assert features (AI message integration, high test accuracy, weekly/monthly summaries) while other files explicitly state LLM and message integration are 'not implemented' or 'TODO'. This mismatch between claimed capabilities and included artifacts is a coherence concern.
Instruction Scope
SKILL.md gives concrete Bash pseudocode to classify content, call an LLM, and write markdown files under memory/just-note/*.md. The runtime instructions don't instruct reading unrelated system config or secrets, and saved files are scoped under memory/just-note. However the instructions reference OpenClaw gateway hooks and an LLM calling mechanism without providing the actual integration code; the provided LLM-calling pseudocode (call_llm / sessions_send / curl) is ambiguous. The skill also uses unspecified variables (e.g., $SOURCE) which are not declared, so runtime behavior may be inconsistent.
Install Mechanism
There is no install spec (instruction-only), so nothing arbitrary is downloaded during install. The only included script is publish.sh which is for publishing and calls the clawhub CLI. No external download URLs or extract operations are present in the provided files.
Credentials
The skill declares no required environment variables or credentials, which matches the fact it intends to store notes locally. However some runtime snippets refer to $SOURCE and to potential LLM API endpoints or tokens (in TODO notes) but do not declare or require them. If the author later wires in direct LLM API calls or message connectors, additional credentials would be required; currently that is not present or declared.
Persistence & Privilege
The skill does not request always:true and is user-invocable. It will write files to a local 'memory/just-note' directory per its instructions, which is expected for a note-taking tool. There is no evidence it attempts to modify other skills or global agent settings.
What to consider before installing
Plain-language checklist before installing or running this skill: - Missing CLI: The package.json and docs expect a bin/just-note executable, but that file is not included here. Do not assume the CLI exists — ask the author for the actual executable or inspect the 'bin/just-note' script before running. - LLM & messaging not implemented: The SKILL.md contains pseudocode for calling an LLM and routing WeChat/Feishu messages, but multiple files state LLM/message integration is TODO. Confirm how the skill will call LLMs (OpenClaw sessions_send, a local CLI, or a direct API) and which credentials (if any) it will require before enabling message auto-ingest. - Docs conflict: Some test reports claim 100% AI accuracy and completed features while other files explicitly list them as unimplemented. Treat the reported test results with caution; ask for a reproducible test or demo. - Filesystem access: The skill uses Bash/Read/Write/Grep. That allows it to read and write files — intended for storing notes under memory/just-note, but verify the implementation to ensure it does not read or exfiltrate unrelated files. - publish.sh behavior: The included publish script calls the clawhub CLI and requires the user to be logged in; publish.sh is for distribution and is not required for runtime, but read it before running. Recommended steps: 1) Ask the author for the missing bin/just-note executable or the real implementation of the CLI. 2) Request clarification / code showing how LLM calls and message routing will be performed and which credentials (if any) will be needed. 3) Run the skill in an isolated test environment (sandbox or container) and inspect what files it creates/writes. 4) If you plan to enable automatic message ingestion, verify the integration path and scope of permissions for any connectors (WeChat/Feishu) before granting them. Confidence is medium because the skill appears honest in intent but the inconsistent/missing artifacts and contradictory test claims prevent a stronger 'benign' verdict.

Like a lobster shell, security has layers — review code before you run it.

aivk978c7k0avg3t1a4xwe4dssmp183mkmmknowledgevk978c7k0avg3t1a4xwe4dssmp183mkmmlatestvk978c7k0avg3t1a4xwe4dssmp183mkmmnotevk978c7k0avg3t1a4xwe4dssmp183mkmmproductivityvk978c7k0avg3t1a4xwe4dssmp183mkmm
97downloads
0stars
1versions
Updated 1mo ago
v0.2.0
MIT-0

记一下 (just-note) - 零摩擦知识记录

一句话定位

像发消息一样记录一切,AI 自动分类整理,让知识自然生长。

核心能力

能力说明
零摩擦输入微信/飞书发消息即可记录,无需打开任何界面
AI 自动分类自动识别 9 类内容(inspiration/idea/knowledge/expense/income/diary/task/quote/other)
AI 标签生成自动生成 3-5 个标签,便于检索
AI 标题生成自动生成简洁标题
统一存储所有记录存储在统一位置,支持多视图呈现
智能检索关键词搜索 + 类型筛选 + 语义检索
日记视图按天聚合记录,支持每日汇总
周报/月报AI 自动生成的周期性总结

内容类型

类型说明示例
inspiration灵感、创意、想法「这个产品功能可以这样做...」
idea想法、读书心得「今天读到的一句话很有启发...」
knowledge知识点、解释「Python 的装饰器原理是...」
expense支出记录「花了¥200 买书」
income收入记录「收到稿费¥5000」
diary日记、感受「今天遇到了一个有趣的人...」
task待办事项「记得下周约医生」
quote引用、名人名言「XXX 说:...」
other其他无法分类的内容-

快速开始

方式 1:微信/飞书消息(推荐)

发送消息即可,AI 自动分类

这个产品功能可以这样做:用户发消息后,AI 自动分类并存入知识库,比手动标签简单多了

AI 自动处理:

  • 类型:inspiration
  • 标签:[product, ai, automation]
  • 标题:产品自动化分类功能灵感
  • 存储:memory/just-note/2026-03/2026-03-26-120000.md

再试一笔支出

花了 200 块买书

AI 自动处理:

  • 类型:expense
  • 金额:200
  • 标签:[book, learning]
  • 标题:买书支出¥200

方式 2:CLI 命令(调试用)

明确指定参数,CLI 直接执行

just-note write --type expense --amount 200 --tags "book,learning" --content "买书"

查看今日记录

just-note today

输出:

## 2026-03-26 今日记录

共 5 条记录:
- inspiration: 2 条
- expense: 1 条 (¥200)
- task: 1 条
- diary: 1 条

[详细列表...]

检索历史记录

按关键词:

just-note search "产品"

按类型:

just-note list --type expense

按日期范围:

just-note list --from 2026-03-01 --to 2026-03-31

笔记格式标准

采用 memory-notes 格式,支持知识图谱:

---
title: "AI 生成的标题"
type: inspiration  # 9 类之一
created: 2026-03-26T12:00:00+08:00
day-id: 2026-03-26  # 用于按天聚合
tags: [tag1, tag2, tag3]
amount: 200  # 可选,收支类型时有
currency: CNY  # 可选
source: wechat  # wechat/feishu/voice/image
---

# AI 生成的标题

## 原始内容
用户发送的原始消息内容...

## AI 整理
- [insight] AI 提取的核心观点 1
- [insight] AI 提取的核心观点 2
- [meta] 金额:¥200(如果是收支类型)

## 关联笔记
- relates_to [[相关笔记标题]]

目录结构

memory/just-note/
├── 2026-03/
│   ├── 2026-03-26-120000.md
│   ├── 2026-03-26-140000.md
│   └── ...
├── 2026-04/
│   └── ...
└── index.json  # 可选,加速检索

命令参考

记录类

命令说明
just-note record "内容"手动记录一条
just-note quick "内容"快速记录(最小化处理)

查询类

命令说明
just-note today查看今日记录
just-note yesterday查看昨日记录
just-note list列出所有记录
just-note list --type <type>按类型筛选
just-note list --from <date> --to <date>按日期范围
just-note search "<keyword>"关键词搜索
just-note diary --date <date>日记视图(按天聚合)

统计类

命令说明
just-note stats总体统计
just-note stats --type expense按类型统计
just-note weekly本周统计
just-note monthly本月统计

导出类

命令说明
just-note export --format flomo导出为 flomo 格式
just-note export --format obsidian导出为 Obsidian 格式
just-note export --format excel导出为 Excel(收支专用)

AI 分类实现

核心理念:AI 做大脑,CLI 做手脚

角色职责特点
AI(我)理解、分类、推理、生成标签有"大脑",会思考
CLI执行明确指令、写入文件无"大脑",纯工具

工作流程

消息模式(主要)

用户微信消息
    ↓
OpenClaw Gateway 接收
    ↓
AI(我)理解内容 → 自动分类、生成标签、提取金额
    ↓
AI 调用 CLI:just-note write --type xxx --tags xxx ...
    ↓
CLI 执行写入 → 文件保存

关键:AI 负责理解,CLI 负责执行。

CLI 模式(备用)

用户明确参数:just-note write --type expense --amount 200 ...
    ↓
CLI 直接执行 → 文件保存
    ↓
不做任何分类/理解

关键:CLI 不做思考,只执行明确指令。

为什么这样设计?

  1. 效率最高 - AI 做擅长的事(理解),CLI 做擅长的事(执行)
  2. 职责清晰 - AI 有"大脑",CLI 是"手脚"
  3. 符合直觉 - 用户发消息 = 让 AI 处理;用 CLI = 自己明确指定

Prompt 模板:

你是一个知识记录分类助手。请分析用户输入的内容,完成以下任务:

1. 识别内容类型(9 选 1):
   - inspiration: 灵感、创意、想法
   - idea: 想法、读书心得
   - knowledge: 知识点、解释
   - expense: 支出记录
   - income: 收入记录
   - diary: 日记、感受
   - task: 待办事项
   - quote: 引用、名人名言
   - other: 其他无法分类的内容

2. 生成 3-5 个标签(简洁、有意义)

3. 生成一个简洁的标题(10-20 字)

4. 如果是收支类型,提取金额和货币单位

5. 提取 1-3 个核心观点/事实(用于 Observations)

输出 JSON 格式:
{
  "type": "inspiration",
  "title": "标题",
  "tags": ["tag1", "tag2", "tag3"],
  "amount": null,
  "currency": null,
  "observations": ["观点 1", "观点 2"]
}

实现细节

1. 消息接收

通过 OpenClaw Gateway 接收微信/飞书消息:

// 伪代码
onMessage(async (message) => {
  if (message.source === 'wechat' || message.source === 'feishu') {
    await processRecord(message.content, message.source);
  }
});

2. AI 分类

调用 LLM 进行自动分类:

# 伪代码
ai_classify() {
  content="$1"
  prompt=$(cat <<EOF
[分类 Prompt 见上方]
用户输入:$content
EOF
)
  response=$(call_llm "$prompt")
  echo "$response"
}

3. 文件写入

生成 Markdown 文件:

write_note() {
  type="$1"
  title="$2"
  content="$3"
  tags="$4"
  observations="$5"
  
  timestamp=$(date +%Y-%m-%d-%H%M%S)
  month=$(date +%Y-%m)
  file="memory/just-note/$month/$timestamp.md"
  
  cat > "$file" <<EOF
---
title: "$title"
type: $type
created: $(date -Iseconds)
day-id: $(date +%Y-%m-%d)
tags: [$tags]
source: $SOURCE
---

# $title

## 原始内容
$content

## AI 整理
$(format_observations "$observations")

## 关联笔记
$(format_relations "$relations")
EOF
}

4. 检索实现

关键词搜索:

search_notes() {
  keyword="$1"
  grep -r "$keyword" memory/just-note/ --include="*.md"
}

按类型筛选:

list_by_type() {
  type="$1"
  grep -l "^type: $type$" memory/just-note/*/*.md
}

5. 日记视图

按天聚合:

diary_view() {
  date="$1"
  echo "# $date 日记"
  echo ""
  echo "## 今日概览"
  
  count=$(grep -l "^day-id: $date$" memory/just-note/*/*.md | wc -l)
  echo "共 $count 条记录"
  
  # 按类型统计
  for type in inspiration idea knowledge expense income diary task quote other; do
    type_count=$(grep -l "^type: $type$" memory/just-note/*/*.md | grep "$date" | wc -l)
    if [ $type_count -gt 0 ]; then
      echo "- $type: $type_count 条"
    fi
  done
  
  echo ""
  echo "## 详细记录"
  grep -l "^day-id: $date$" memory/just-note/*/*.md | while read file; do
    cat "$file"
    echo ""
    echo "---"
    echo ""
  done
}

配置

环境变量

变量说明默认值
JUST_NOTE_STORAGE存储路径memory/just-note
JUST_NOTE_LLM_MODELLLM 模型qwen3.5-plus
JUST_NOTE_AUTO_TAG是否自动标签true
JUST_NOTE_AUTO_RELATE是否自动关联true

配置文件

~/.just-note/config.yaml:

storage: memory/just-note
llm:
  model: qwen3.5-plus
  temperature: 0.3
features:
  auto_tag: true
  auto_relate: true
  daily_summary: true
  weekly_report: true
notifications:
  daily_summary_time: "21:00"
  weekly_report_time: "Sunday 20:00"

最佳实践

1. 记录技巧

  • 简短为好 - 一条记录 1-3 句话最佳
  • 及时记录 - 灵感来了立刻记
  • 不用整理 - AI 会自动分类标签
  • 定期回顾 - 每周/每月查看汇总

2. 检索技巧

  • 用标签搜 - just-note search "#product"
  • 按类型筛 - just-note list --type expense
  • 按日期找 - just-note diary --date 2026-03-26

3. 复盘技巧

  • 每日回顾 - 睡前查看今日记录
  • 每周汇总 - 周末查看周报
  • 每月总结 - 月末查看月报

与 flomo 对比

维度flomojust-note
输入方式微信/APP微信/飞书
整理方式手动标签AI 自动分类
内容类型通用笔记9 类(含收支)
数据位置云端本地 + 云端可选
检索方式标签 + 关键词关键词 + 类型 + 语义
复盘功能每日回顾每日/每周/每月 AI 总结
收支统计
价格¥12/月免费

未来规划

阶段 1(MVP)- 1-2 周

  • 核心记录功能
  • AI 自动分类
  • 基础检索
  • 每日汇总

阶段 2(增强)- 2-4 周

  • 语义检索
  • 智能关联
  • 收支统计图表
  • 导出功能

阶段 3(产品化)- 1-2 月

  • Web 界面
  • 多端同步
  • API 开放
  • 插件系统

常见问题

Q: 和 flomo 比有什么优势?

A:

  1. AI 自动分类,不用手动标签
  2. 支持收支记录和统计
  3. 数据存储在本地,完全自主
  4. 免费开源

Q: 数据会丢失吗?

A:

  • 数据存储在本地 memory/just-note/ 目录
  • 建议定期 Git 备份或同步到云端
  • 可以导出为 flomo/Obsidian 格式

Q: 分类不准确怎么办?

A:

  • 可以手动修改笔记的 type 字段
  • 分类准确率会随着使用提升
  • 欢迎反馈改进建议

许可证

MIT License

贡献

欢迎提交 Issue 和 PR!

GitHub: https://github.com/your-org/just-note

Comments

Loading comments...