WeChat Read’s official API offers six calls. This offers forty.

What it does

weread-omni wraps WeChat Read (微信读书) in a single TypeScript SDK and JSON-first CLI, exposing forty read-write operations. It covers the official API’s six read-only endpoints—shelf queries, search, stats—and keeps going into territory Tencent hasn’t opened: writing notes and reviews, importing personal EPUBs and PDFs, subscribing to WeChat Official Accounts (公众号), and even querying the built-in AI. A single implementation powers the CLI, the SDK, and an installable agent skill.

The interesting bit

The project treats the CLI as an agent-native interface: every command emits JSON, and the bundled skill teaches agents how to page through results and ask for confirmation before mutating data. It also caches content locally in SQLite, so repeated lookups hit disk instead of the server.

Key highlights

• 40 operations versus the official skill’s 6, including write access to highlights, reviews, and bookshelf state
• Imports personal books in EPUB, PDF, MOBI, TXT, and AZW3 formats
• Full WeChat Official Account support: subscribe, list articles, export feeds to RSS/Atom/JSON
• Multi-account support with alias-based switching and locked-down local config permissions (0700/0600)
• Local content library with deduplicated storage and offline replay via SQLite

Caveats

• Unofficial and unaffiliated with Tencent; requires a real WeChat account and QR-code login
• Local content library stores reading material in plaintext and requires SQLite WAL support—file systems lacking WAL trigger a warning and disable caching
• Version 0.1.0 briefly shipped under the weread command name, colliding with the official skill

Verdict

Worth it if you are scripting WeChat Read workflows—archiving public accounts, bulk-importing books, or syncing notes. Stick to the official skill if you only need occasional read-only queries and prefer staying inside the guardrails.