WeChat Intelligence Hub skill
A local-first set of Codex skills and a read-only CLI that turns your own WeChat history into search, digests, follow-ups and opportunity tracking
- 2,795 ★ on the repository
- Last commit
- AGPL-3.0
Broad access
Why this level
- Reads private chats and the local WeChat database and handles encryption keys
- The experimental access helper may restart and re-sign a copy of WeChat
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/wechat-intelligence-hubDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Reads private chats and the local WeChat database and handles encryption keys
- The experimental access helper may restart and re-sign a copy of WeChat
This is third-party code. Review the repository files before installing.
What it does
The project builds a personal database from local WeChat history that the agent searches, reports on and uses to remind you of tasks. The repository ships two Codex skills and a read-only CLI: wechat-cli is the single entry to its own database reader, and wechat-intelligence-hub turns records into daily briefs, reply queues, unkept promises, opportunities and reactivation leads over any chosen period. Work is strictly read-only: the skill does not send messages, reply, add contacts or drive the WeChat interface, and reply drafts stay local. The reader reads a database the user has authorized, while a separate experimental access helper can prepare read materials only with explicit confirmation and the user's own system authorization. All data and reports stay on the user's machine, and keys and chats must not be sent to the maintainer or a community.
Who it is for. For people who run deals and clients in WeChat: sellers, founders and creators who need to make sense of their own chats.
Good fit when
- You need a digest of your own WeChat history for a day, a week or a chosen period
- You need to search all chats for what was discussed about a person, topic or project
- You need to track promises, reply prompts and business opportunities
Not a fit when
- You have no access to your own local WeChat database and read materials
- You need to send or change WeChat messages rather than only read
- You want access to someone else's chats; the project works only with your own account
Example request
Build a WeChat brief for the last 24 hours: groups, key contacts, what needs a reply and business opportunitiesLimitations
It works through Codex and requires an accessible local WeChat database and authorized read materials, covering only data already synced to the machine. Database compatibility depends on the WeChat version, and notification mode is inbound previews only, not the full history. First access is an experimental optional step: it may restart WeChat and re-sign a copy, requires explicit confirmation and the user's authorization, and the path on new macOS machines is not guaranteed. Full reading needs SQLCipher and Zstandard. The license is AGPL-3.0, and closed integration needs a separate commercial license. The README also mentions a separately paid community and author services, which the tool itself does not require.
How to disable. Remove the skills/wechat-cli and skills/wechat-intelligence-hub folders from the Codex skills directory along with the local engine installed there. Remove the local database, profile and reports manually, since they are stored only on your machine.
FAQ
Does it read someone else's chats?
No. The project works only with your own account on your own machine, strictly read-only, and does not send, change or delete messages.
Does installation immediately read the whole history?
No. A successful install is not the same as chat access. Full reading needs an authorized local database and access materials, and the reader does not obtain keys on its own.
Discussion
No questions or reviews yet.
Related
SkillsA Codex skill that finds potential first customers from public demand signals, verifies an open contact route and builds a report.
Six skills for Codex run TikTok product selection, account audits, lead-generation content and ads reviews, driven by data, not guesswork.
49 Chinese-language skills for private-domain sales, business models, content and AI rollout, with a 28,000-note knowledge base.
ChatbotX MCP server: contacts, conversations, flows, broadcasts and automation in WhatsApp, Instagram and other channels
A skill that searches 15 channels for creators with their own audience, gathers public contacts and appends them to a Google Sheet
Two LinkedIn skills via Linked API: people and company search, messages, posts and a lead pipeline with connection invites.
README in short
The README describes WeChat Intelligence Hub as a standalone project rather than a prompt bundle: it turns local WeChat history into a verifiable, actionable personal database with contact history, group topics, a reply queue, promises, business opportunities and reactivation leads. One product is split into two Codex skills: the dependency layer wechat-cli with a read-only reader and the user entry wechat-intelligence-hub with reports and rules. Installation goes through git clone and install.sh or a direct instruction to Codex, and full reading needs an authorized local database, SQLCipher and Zstandard. A separate experimental access helper with explicit confirmation can prepare read materials, while the reader does not obtain keys. Privacy is called out: read-only, data never leaves the machine, and repository examples are fictional. The license is AGPL-3.0 with an optional commercial license.
README badge
Original SKILL.md
--- name: wechat-intelligence-hub description: WeChat Intelligence Hub（微信个人情报库），本地只读微信情报 Skill。用于按任意明确时间范围，或围绕某条信息、某个人、某个群、某个产品/物品、某个微信标签、某个项目或事件检索和总结微信聊天；也用于生成群聊与私聊报告、发现并跟进商单/培训/咨询/项目合作、聚合跨群链接、检查待回复和待兑现承诺、维护商机状态及筛选值得复联的人。 --- # 微信个人情报库 ## Mission 把本地微信记录转化为可核查、可执行的个人情报。优先回答：发生了什么、和当前用户有什么关系、现在该做什么、依据是哪条消息。不要只生成更长的摘要。 ## Non-Negotiables - 只读微信。不得发送、回复、转发、加好友或自动操作微信；回复内容只能作为本地草稿。 - 原始聊天、联系人和未公开商业信息只留在本机。不要发送给外部搜索或云端研究服务。 - 以原消息为证据。区分 `已确认`、`高概率`、`待核实`，不得把转发者自动认定为品牌方或组织者。 - 最新、现在、刚刚、今日等请求必须先刷新相关数据。索引过旧或采集失败时，明确说明覆盖缺口。 - 用户未给时间时默认过去 24 小时；用户给出时间范围时严格使用该范围。