This lesson configures a local `wechat-manager` Skill so WorkBuddy can read, search, summarize, analyze, and export WeChat conversations through natural language.
The course now provides one Windows + macOS auto-detect prompt. It checks the current system and usable local configuration before choosing a platform-specific read-only route; never mix the two platforms' low-level steps manually.
Before you start | Third-party solution, account risk, and responsibility
This course uses a third-party open-source solution, not a Tencent or WeChat official feature. Reading, analyzing, or exporting local WeChat data or a running WeChat process may trigger account anomalies, restrictions, or WeChat account disablement. Platform rules, client implementations, and tools may change at any time.
This material is for learning and teaching only. Test only on your own computer, your own test account, and the smallest possible data scope. The course and site accept no responsibility for account, data, business, privacy, or other loss resulting from its use. Never upload or share databases, keys, masterkeys, snapshots, passwords, or complete chat histories.
How to use this lesson
The bottom of this page has one Windows + macOS auto-detect Markdown download. Give it to WorkBuddy or another AI tool; it first checks the system, WeChat version, login state, actual Skills root, and usable local tools before choosing a platform-specific route. Read its read-only, privacy, batch-message, and stop-on-failure rules in full before acting.
Use the order: check the environment, reuse a working existing configuration or complete the first installation check, create or read a local snapshot, verify a real session or group, then enable search, summaries, and exports. Do not repeat initialization, expose keys, or switch to unverified tools when a step fails.
Windows-specific notes
The Windows route has been validated on Windows 10 and WeChat 4.1.10.27: Config.Cipher scanning, HMAC validation, and local decrypted snapshots, without downgrading WeChat, using a debugger, DLL injection, or hooks. The unified prompt chooses this tested course route when it detects Windows.
That result covers only the stated test environment, not every Windows or future WeChat version. In your own test environment, use the file's `doctor → groups → history → search` acceptance sequence. It is successful only after real messages and search results are returned.
macOS must follow version-specific paths
macOS must not copy Windows low-level key-extraction steps. The unified prompt first checks for a working historical `wechat-cli` configuration: if real queries work, it reuses it without initialization. If no configuration exists or it has failed, it makes one original `wechat-cli` initialization attempt. `0 keys` stops that legacy path; it is not a successful Skill setup.
A newer-macOS compatibility branch using `wcdb-key-tool` is considered only with explicit user approval; it may involve Xcode Command Line Tools, re-login, `sudo`, LLDB, or WeChat.app re-signing, so it never runs automatically. For network, certificate, DNS, or permission errors, use the one-time npm mirror/official-registry fallback and macOS Full Disk Access checks. Do not disable TLS verification, use long-term `sudo npm install -g` workarounds, or install identically named tools from untrusted mirrors, drives, or binaries.