Link Saver AI

• You regularly come across articles, blog posts, or resources you want to read later but keep forgetting about.
• You want quick AI summaries of saved links so you can triage your reading list without opening every tab.
• You prefer managing your reading list through natural conversation — in English, Chinese, or both — rather than clicking through a separate app.
• You want a weekly digest that surfaces your unread links, highlights top topics, and suggests what to read next.

• You need to sync your reading list across multiple devices or share it with other users — this skill stores data locally on a single machine.
• You require offline content archival with guaranteed full-text retrieval; content fetching depends on live web access and may fail for paywalled or restricted pages.

When you save a link, the skill fetches the page content and writes a 2–3 sentence summary capturing the main point and significance. It also assigns 1–3 tags from a built-in taxonomy (e.g., tech, AI, finance, tutorial) or creates new tags when none fit.

Trigger any action by typing conversationally — "save this link", "帮我存这个链接", "what was that article about AI?", or "show my unread links". The skill mirrors your language in its response.

Search your saved links by tag, title substring, domain, or free-form query. When no exact match is found, the skill applies AI reasoning to find the most semantically relevant links from your stored summaries.

Ask for a detailed summary of any saved link to get a structured breakdown: a TL;DR, key bullet points, notable quotes or data, and a brief AI take on the article's significance. The full content is cached locally for faster repeat access.

The /link_digest command generates a weekly summary showing how many links you saved and read, your top tags, and a suggested next read chosen for recency and topic diversity. Delivery history is tracked so you're warned if digests stop arriving.

The skill works out of the box with no external account or API credential setup. All link data is stored locally in ~/.openclaw/workspace/links/ as plain JSON files.

Paste a URL into the chat with "save this" and the skill fetches the page, writes a summary, and tags it automatically. You get a confirmation with the title, tags, and summary — and a running count of your total and unread links.

Type "show my unread links" or /reading to get a sorted list of everything you haven't read yet, each with its summary and tags, so you can quickly decide what to open next.

Say "what was that article about AI agents?" or use /link_search with a keyword. The skill searches titles, tags, summaries, and domains — and falls back to semantic matching if nothing is an exact fit.

Run /link_digest on a Friday to see everything you bookmarked that week, your read/unread ratio, the topics you gravitated toward, and a single recommended article to start with.