Skill flagged — suspicious patterns detected

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

Family Soul

v0.1.0

从家庭群聊记录(微信/WhatsApp/其他)提炼数字人格。 输出 soul.md(集体人格)+ 每位成员的 persona 文件,可直接用于 AI agent 的人格底座。 关键词:群聊分析、家庭人格、soul、persona、数字人格、聊天记录、微信导出、人格提炼。

0· 109·0 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 zengury/family-soul.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "Family Soul" (zengury/family-soul) from ClawHub.
Skill page: https://clawhub.ai/zengury/family-soul
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 family-soul

ClawHub CLI

Package manager switcher

npx clawhub@latest install family-soul
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
high confidence
Purpose & Capability
The name/description (family chat -> persona/soul.md) aligns with the included parsing/denoising/extraction/synthesis pipeline. Having multiple extractor implementations (Anthropic, Kimi/Moonshot, OpenClaw variants) can be legitimate as provider/backend choices, but the repository does not declare those alternate credentials in SKILL.md or metadata and instead embeds provider-specific code paths and keys. That mismatch is unexpected.
!
Instruction Scope
SKILL.md instructs the agent to run scripts that parse user-provided chat files and send chunked text to external LLM APIs. That core scope is appropriate for the stated purpose, but the runtime instructions require ANTHROPIC_API_KEY while several pipeline scripts will call Kimi/Moonshot endpoints (and some contain hard-coded API keys/fallbacks). Those scripts therefore expand the instruction surface to send sensitive chat contents to additional third parties that the SKILL.md does not mention.
Install Mechanism
This is an instruction-plus-code skill (no installer). There is no external download during install (low install risk), but the package includes runnable Python scripts which will execute and perform network calls. Because code is shipped with the skill, review of the files is required prior to execution.
!
Credentials
SKILL.md and README state ANTHROPIC_API_KEY (Anthropic) is required, but multiple pipeline scripts reference KIMI_API_KEY / MOONSHOT_API_KEY and three different scripts include an embedded API key string (e.g. 'sk-kimi-Sgsy7YYJ...'). Hard-coded keys and implicit fallbacks (env var OR hardcoded value) are disproportionate for the stated purpose and risk unauthorized transmission of private chat data to third-party services. The skill metadata declares no required env vars, which is inconsistent with runtime requirements.
Persistence & Privilege
The skill does not request always:true and doesn't declare system-wide modifications. It will read user-provided files and write outputs to a local output directory; that is expected behavior for this functionality.
Scan Findings in Context
[hardcoded_api_key] unexpected: Multiple files contain an apparent hard-coded API key for a Kimi/OpenClaw endpoint (e.g. pipeline/03_extract_kimi_openclaw.py, pipeline/03_extract_kimi_v2.py, pipeline/03_extract_simple.py). Some scripts use an env var OR fall back to this literal key, meaning if the env var is missing the code will still send data using the embedded credential. This is unexpected and risky for a user-facing skill that processes private chat exports.
[undeclared_external_providers] unexpected: SKILL.md documents Anthropic use and asks the user to set ANTHROPIC_API_KEY, but multiple alternative provider clients (Kimi/Moonshot/OpenAI-compatible wrappers) are present. Those alternate providers are not declared in the skill manifest or SKILL.md as supported/backends and some are triggered automatically by code paths.
[sample_chat_data_in_repo] expected: The repository includes sample/raw exported group chat data (data/raw/群聊_...json). Having example data is reasonable for a demo, but it contains realistic personal messages and metadata; users should avoid running the skill on real data without inspection and privacy considerations.
What to consider before installing
Key things to check before installing or running this skill: - Do not run on real family chat exports until you are comfortable with where the data goes. The code will send chat text to external LLM APIs. - Inspect scripts/run_forge.py (entrypoint) to confirm which extraction implementation it chooses by default. If it calls any of the 'kimi' or 'openclaw' variants, those scripts include an embedded API key and/or will call third-party endpoints. - Search the repository for the literal string 'sk-kimi-' and for domains like 'kimi.com' or 'moonshot.cn'. If present, remove or replace hard-coded keys and change any fallback logic so the code requires an explicit environment variable (do not silently fall back to an embedded credential). - If you intend to use only Anthropic/your own key: ensure all scripts invoked by run_forge.py rely only on ANTHROPIC_API_KEY and fail explicitly if it is missing. Consider removing unused provider scripts from the skill bundle. - Run the pipeline in an isolated environment (network-restricted or monitored) the first time, and test with synthetic/non-sensitive data to confirm behavior and endpoints. - Because the repository includes sample chat exports with PII, delete or replace those files before adding your own real exports. - If you are not comfortable auditing Python code, do not install/run this skill. At minimum, ask the author to remove embedded credentials, to document all external endpoints/providers, and to require explicit env vars for any third-party APIs.

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

latestvk979txy93eqzkef7mdatet765n83hjnw
109downloads
0stars
1versions
Updated 1mo ago
v0.1.0
MIT-0

SKILL: Soul Forge — 家庭数字人格提炼

这个 skill 把一份家庭群聊记录变成可用于 AI agent 的人格文件。 基于数字民族志方法论:用 AI 完成「田野调查」→「人格合成」的完整流程。


触发条件

以下情况触发此 skill:

  • 用户说"帮我分析聊天记录"、"生成 soul 文件"、"提炼家庭人格"
  • 用户提供了 .json 聊天导出文件
  • 用户说"运行 soul-forge"、"开始人格提炼"
  • 用户问"怎么用聊天记录生成 persona"

执行流程

第一步:确认输入

询问用户:

  1. 聊天记录文件路径(支持微信 WeFlow 导出的 JSON 格式)
  2. 输出目录(默认:~/soul-forge-output/
  3. 家庭成员角色配置(默认:dad/mom/child 三人结构)

确认 ANTHROPIC_API_KEY 已设置(需要调用 Claude API)。

第二步:后台运行 pipeline

调用:

python3 {SKILL_DIR}/scripts/run_forge.py --file {用户提供的文件路径}

四个阶段,agent 依次推进:

阶段脚本说明预计时间
101_parse.py解析原始聊天 JSON → 标准化消息30秒
202_denoise.py去噪、按时间分块1分钟
303_extract.pyClaude Haiku 批量提取行为模式(Batches API)10-30分钟
404_synthesize.pyClaude Opus 综合生成 soul.md + persona5-15分钟

阶段3说明:使用 Batches API 异步处理,成本低,自动缓存进度。 如被中断可用 --resume 恢复,不重复计费。

第三步:进度汇报

解析 run_forge.py 的标记输出:

  • [STAGE:N:START] → 告知用户"正在进行阶段N"
  • [STAGE:N:DONE] → 告知用户"阶段N完成"
  • [PROGRESS:N/M] → 展示进度条
  • [OUTPUT:path] → 列出生成的文件
  • [ERROR:msg] → 报告错误,建议用户如何处理
  • [DONE] → 宣布完成,展示所有输出文件

第四步:完成后

输出文件说明:

soul-forge-output/
├── soul.md          ← 集体人格,可直接作为 AI agent SOUL.md 使用
├── persona_dad.md   ← 爸爸个人人格
├── persona_mom.md   ← 妈妈个人人格
└── persona_child.md ← 孩子/子女人格

询问用户是否要:

  • 将 soul.md 安装为当前 agent 的 SOUL.md
  • 为每个 persona 创建独立 agent

进阶用法

只更新 soul,不重新生成 persona

告诉 agent:「soul-forge 只更新 soul,跳过 persona」

内部:python3 run_forge.py --file {path} --soul-only

只重新生成 persona(soul 已存在)

告诉 agent:「soul-forge 只刷新 persona」

内部:python3 run_forge.py --file {path} --persona-only

从中断处恢复

告诉 agent:「soul-forge 继续上次的任务」

内部:python3 run_forge.py --resume

查看当前进度

告诉 agent:「soul-forge 状态」

内部:python3 run_forge.py --status


支持的输入格式

格式来源说明
微信 WeFlow JSONWeFlow 工具导出完整支持
标准 CSV自定义导出需包含 sender/timestamp/content 列

微信导出方法:用 WeFlow(Mac)→ 选群聊 → 导出 JSON 格式。


成本估算

一份 2-3 年的家庭群聊(~500 块对话):

  • 阶段3(Haiku Batches):约 $0.5-1.0
  • 阶段4(Opus):约 $2-5
  • 合计约 $3-6,一次性

常见问题

Q: 阶段3 很慢怎么办? A: Batches API 通常 10-30 分钟,这是正常的。agent 会持续轮询状态,不需要人工干预。

Q: 中途断了怎么办? A: 说「soul-forge 继续」,脚本会从断点恢复,已完成的阶段不会重复执行。

Q: API key 在哪里设置? A: export ANTHROPIC_API_KEY='sk-ant-...',或在 OpenClaw 的环境变量设置里配置。

Q: 支持几个人的群聊? A: 默认三人(dad/mom/child),可在 pipeline/config.py 修改角色配置。


方法论背景

基于数字民族志(Digital Ethnography):

  • 阶段1-2:田野记录整理(去噪、结构化)
  • 阶段3:系统性观察(Haiku 提取五维度行为模式)
  • 阶段4:民族志分析(Opus 综合「厚描」)

Clifford Geertz:「浅描记录行为,厚描解释意义。」

soul.md 是厚描的产物——不是行为清单,而是理解这个家庭需要什么样的解释框架。

Comments

Loading comments...