WeRead Assistant

Background: The Challenge of Automating WeRead Data Management

WeRead, a popular reading app, accumulates rich personal data such as notes, highlights, reading time, and preferences. However, its official client lacks an open API, making it difficult for users to automatically extract or analyze this data. Manual efforts are time-consuming, error-prone, and cannot enable cross-platform integration—for example, generating an annual reading summary or batch-viewing book notes is limited to in-app features without programmatic access.

Core Capabilities and Workflow of the Skill

This skill leverages an Agent API Gateway to call WeRead's private interfaces, enabling programmatic data access. Its capabilities span the full reading lifecycle:
- Book Search and Info Retrieval: Use /store/search to find books by keyword, returning bookId and other details; fetch chapter directories and reading progress via /book/detail.
- Shelf and Reading Statistics: Manage the bookshelf with /shelf/sync, supporting albums/audiobooks (calculated as books.length + albums.length + (mp非空 ? 1 : 0)); generate reading summaries including time, days, and preference analysis through /readdata/summary.
- Notes and Review Management: View highlights and thoughts via /book/bookmarklist, retrieve public reviews with /book/readreviews, and analyze popular highlight trends.
- Personalized Recommendations: Discover books through /discover/recommendations, expanding reading exploration.

Key Workflow Steps:
1. Authentication Setup: Configure the environment variable WEREAD_API_KEY (format wrk-xxxxxxxx); include Authorization: Bearer $WEREAD_API_KEY in request headers, with the API automatically associating user identity.
2. Request Construction: Send POST requests with JSON body containing mandatory skill_version: "1.0.3" and api_name (e.g., /shelf/sync). Business parameters must be flattened at the root level—never nest them in objects like params.
3. Response Processing: Parse the JSON response and check errcode; if upgrade_info appears, immediately pause and upgrade the skill before retrying. Field interpretation must follow dedicated documentation (e.g., readdata.md), not literal translation.
4. Result Presentation规范: Use numbered lists for easy selection; convert timestamps (e.g., updateTime) to YYYY-MM-DD format; transform reading time from seconds to "X hours Y minutes"; for books, chapters, or highlights, construct deep links (e.g., https://weread.qq.com/web/reader/...) to open directly in the app.

Constraints and Applicability

• Data Dependencies: Requires stable internet; API responses depend on WeRead's service status, potentially returning error codes that need user checks or retries.
• Permission Limits: Some interfaces (e.g., note export) may have API permission constraints, preventing full data access; bookshelf counts must strictly follow the documented formula to avoid errors.
• Field Accuracy: Response field names may conflict with intuition (e.g., count); explanation must refer to the corresponding documentation (e.g., shelf.md), avoiding guesses based on experience.
• Upgrade Mechanism: The skill version is reported via skill_version; the server may return upgrade_info to trigger mandatory updates, which—if ignored—could disable functionality.
• Use Cases: Suitable for personal reading data analysis, automated note organization, or integration development, but does not support batch book modifications or direct reading control (e.g., page-turning); limited to query and statistics operations.