小红书笔记转文字 Xiaohongshu note to textv1.1.0

A Xiaohongshu creator or reposter sends a note link (xiaohongshu.com, xhslink.com) to the skill and gets the full caption plus per-image OCR text, so they can repurpose the content into scripts, subtitles or reposts. The returned Markdown keeps image order and key points, saving manual transcription. This is the primary use case listed in the SKILL.md.

Marketing researchers collect high-performing Xiaohongshu notes and need the complete content plus engagement stats (likes, comments, collects, shares, tags) for analysis. The skill returns structured data with stats.as_of and key_points, ideal for building a research dataset or spotting trends. Cached links return at zero credits, making bulk re-reads cheap.

A localization team or e-commerce seller operating on Chinese platforms pulls a Huo-brand note and uses the optional translation flag to produce English (or other language) versions of the caption and image text. The base OCR plus translation add-on (--translate-to en) gives them a publishable localized draft quickly.

An analyst or student saves useful Xiaohongshu notes (e.g., travel guides, recipes, tutorials with text in images) into a personal knowledge base. The skill converts them to Markdown with ordered images and OCR, ready for pasting into Notion, Obsidian or documentation, without needing a Xiaohongshu account or Cookie.

A podcaster or journalist encounters a video-based Xiaohongshu note and needs the spoken transcript and on-screen text. Although the primary purpose is image notes, the same digest endpoint also handles video notes and gives a verbatim script; for full handling they are pointed to the linkdigest-video-transcript skill. This helps turn video content into searchable text.

Users can run the skill without an API key and view one free summary per day. Registering gives 10 free credits (one-time) to unlock full outputs; thereafter reading a 6-image note costs 1 credit (~¥0.14) and larger notes scale up to 3 credits. This drives sign-ups and paid conversions while keeping a free trial available.

LinkDigest sells per-call access to its note-to-text API via HTTP or MCP, charging credits based on output size and optional add-ons like breakdowns or translations. Pricing is transparent and cache hits are free, encouraging repeat API traffic and integrations. The model targets developers and automation builders rather than only human end-users.

The service is exposed as a remote Streamable HTTP MCP server (linkdigest.dev/mcp) with a digest_url tool, allowing Claude Code or other MCP clients to call it directly from an agent workflow. This embeds Xiaohongshu extraction into autonomous agent pipelines, generating recurring API usage. It positions LinkDigest as an infrastructure component in the emerging AI agent ecosystem.