Huashu WeRead

Why it's ranked here

The strongest idea is its refusal to treat a saved book as a read book. Cross-checking shelves, notes, and recent activity gives recommendations a defensible personal basis. Explicit checkpoints, availability checks, fallback behavior, and a single strongest recommendation make the workflow unusually actionable. Its quality still depends on complete account data and heuristic thresholds.

What's good

Each workflow has a clear decision process rather than a loose prompt. Recommendations identify missing schools, time periods, abstraction levels, or neighboring subjects. Learning paths add staged reading, time estimates, self-check questions, and an acceptable stopping point. Note synthesis clusters ideas instead of dumping highlights. Error handling tells users when data is missing, retries API failures once, and avoids silent degradation.

Tradeoffs

It requires the official WeRead skill, an API key, and live access to the WeRead gateway. Note-count thresholds can underrate readers who rarely annotate, while title-keyword matching may classify topics crudely. New or empty accounts lose much of the personalization. Several checkpoints require user input, and review publishing or social-image creation depends on separate skills rather than this package.

How to use it well

Use it when your WeRead account contains enough shelf and note history to reveal patterns. It fits readers choosing a next book, entering a subject systematically, consolidating highlights, or drafting a periodic reading retrospective. State the topic, desired recommendation count, availability preference, or publishing format upfront. Do not use it as a replacement for the official data connector or as a direct publishing system.

Observed

License

MIT, permitting personal and commercial use.

Packaging

Installed through the skills CLI with npx skills add alchaincyf/huashu-weread.

Interface

Prompt-driven agent skill layered over the official WeRead HTTP API gateway.

Dependency

Requires the official WeRead skill and a WEREAD_API_KEY environment variable.

Agent support

README lists Claude Code, Cursor, Codex, OpenClaw, and Hermes.

Repository structure

Four workflow documents, shared methodology documents, one worked example, and a JSON prompt fixture.