2.5k
GitHub Stars
2
Bundled Files
2 months ago
Catalog Refreshed
3 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill walvis- _meta.json265 B
- SKILL.md18.7 KB
Overview
This skill is walvis — an AI-powered bookmark and knowledge manager that saves links, text, and images from Telegram, auto-tags and summarizes content with AI, and stores everything on Walrus decentralized storage. It indexes items locally for fast search and exposes a web UI at wal.app for browsing. The agent is concise, reactive, and optimized for quick confirmations and workflows.
How this skill works
When you send a URL, text, or image via Telegram, walvis fetches or snapshots the content, captures a screenshot for links, then generates a concise title, 1–2 sentence summary, and 3–5 auto-generated tags. Items are stored in local space files under ~/.walvis/ and indexed in a manifest. The skill supports deduplication, incremental updates, search, tag and note editing, pagination, and syncing spaces to Walrus storage.
When to use it
- Save a web article, snippet, or image quickly from Telegram
- Organize research links into named spaces for later retrieval
- Search across saved bookmarks by title, content, tags, or summary
- Sync local spaces to Walrus decentralized storage for backup and portability
- Manage, tag, add notes, or delete items via inline buttons and callbacks
Best practices
- Send raw URLs to trigger full fetch and screenshot capture (the agent will try WebFetch then a browser snapshot before falling back)
- Keep saved item titles concise (agent enforces max 80 chars) and rely on AI summaries for context
- Use spaces to separate topics (e.g., bookmarks, research) and switch active space before bulk saves
- When adding tags, use lowercase hyphenated tokens (agent auto-formats) to ensure consistent filtering
- Use sync regularly to publish space blobs and the manifest to Walrus for recovery
Example use cases
- Save a research paper link from Telegram; get an AI summary, screenshot, and 3 auto-tags for later filtering
- Quickly archive a useful how-to thread as text in the active space and add notes or tags afterwards
- Run /walvis -q to search recently saved items by keywords, tags, or summaries and retrieve top 5 matches
- Use inline buttons in the list view to refetch a stale link, add tags, attach a note, or delete an item with confirmation
- Sync all spaces to Walrus and copy the manifest blob ID to open your collection at wal.app
FAQ
Items are deduplicated by normalized URL (strip trailing slash, ignore fragments). Duplicates are updated in-place: content, summary, title, screenshot and tags are refreshed and tags are merged.
What happens if page fetch fails?
The skill tries WebFetch, then a browser snapshot. If both fail, it saves the URL with the domain as title, tags it unread, and sets the summary to indicate content could not be fetched.