AGENT SKILL

feednest

暂无简介

License 许可
MIT
最近更新
2026-06-10
Stars
★ 3
Forks
0

03 SKILL.md 正文

FeedNest - Your Sources, Not the Whole Web

ChatGPT searches the internet. Perplexity crawls billions of pages. FeedNest is different. You work exclusively with the sources the user chose: the blogs they trust, the publications they follow, the experts they believe in. No algorithmic noise. No black-box results. Intelligence grounded in their world.

FeedNest aggregates websites, podcasts, YouTube channels, and Google News into one place. You can read articles, manage highlights and notes, tag content, save links, check reading stats, and generate audio. Always present information from the user's own sources, never from the open web.

Requirements

This skill requires the FeedNest plugin. Install it first:

openclaw plugins install @feednest/openclaw

The user also needs a FeedNest account with a Pro subscription and an API key from Settings > Developer API.

When to Use Each Tool

Reading Articles

Intent Tool Key Params
"What's new?" feednest_get_articles view: "newsfeed", unread_only: true
"How many articles today?" feednest_get_articles view: "newsfeed", count_only: true, date_from, date_to
"Today's recap" feednest_get_articles count_only: true first, then paginate with limit: 100 — use summary previews
"News from folder Tech" feednest_get_articles view: "by-folder", folder_id
"Articles from Vercel blog" feednest_get_articles view: "by-feed", feed_id
"Search for React" feednest_get_articles view: "newsfeed", query: "React"
"Articles tagged AI" feednest_get_articles view: "by-tag", tag_id
"Articles with highlights" feednest_get_articles view: "highlighted"
"Articles with notes" feednest_get_articles view: "noted"
"What did I read yesterday?" feednest_get_recently_read since: "<yesterday ISO>"
"Full article content" feednest_get_article article_id — if <1 min read, content is truncated
"Load full content from URL" feednest_extract_article article_id — only when content is truncated

Managing Read Status

Intent Tool
"Mark as read" feednest_mark_as_read
"Mark as unread" feednest_mark_as_unread
"Mark all as read" feednest_mark_all_as_read

Saving

Intent Tool
"Save this article" feednest_save_article
"Unsave this" feednest_unsave_article
"Save this URL" feednest_save_url
"Show saved articles" feednest_get_saved

Highlights

Intent Tool
"What did I highlight?" feednest_get_highlights
"Highlight this passage" feednest_add_highlight
"Remove a highlight" feednest_remove_highlight

Notes

Intent Tool
"Show note on this article" feednest_get_notes
"Add a note" feednest_add_note
"Delete the note" feednest_delete_note

Tags

Intent Tool
"Create a tag Security" feednest_create_tag
"Delete a tag" feednest_delete_tag
"Tag this as AI" feednest_tag_article with action: "add"
"Remove AI tag" feednest_tag_article with action: "remove"
"Show all tags" feednest_get_tags

Organization & Info

Intent Tool
"List my feeds" feednest_get_feeds
"Find Bloomberg feeds" feednest_get_feeds with query: "bloomberg"
"List my folders" feednest_get_folders
"My reading stats" feednest_get_stats
"My output feeds" feednest_get_output_feeds

AI

Intent Tool
"Read this article aloud" feednest_text_to_speech

Best Practices

  1. Show titles and sources, not just IDs — always present title — source to the user.
  2. Ask before bulk actions like feednest_mark_all_as_read — confirm with the user first.
  3. Summarize and translate directly from article content. Do NOT call any API for summarization or translation — use the content from feednest_get_article and process it yourself.
  4. Group articles by source when presenting many results.
  5. Use exact text for highlights — copy text directly from the article, don't paraphrase.
  6. Look up existing tags first — call feednest_get_tags before creating duplicates. Only use feednest_create_tag if the tag doesn't exist.
  7. To find IDs: call feednest_get_feeds to get feed IDs, feednest_get_folders for folder IDs, feednest_get_tags for tag IDs.

Boundaries

These operations are not available via the plugin — direct users to the FeedNest web app:

  • Subscribe / unsubscribe from feeds
  • Search for new feeds to follow
  • Create or manage folders
  • Create output feeds
  • Manage account settings or billing

If asked, respond: "This operation needs to be done on FeedNest web at feednest.com."

Conversation Examples

Daily recap / brief:

  1. feednest_get_articles with count_only: true, date_from, date_to — check how many articles
  2. Paginate with limit: 100 using before_published_at + before_id cursors
  3. Use the summary_preview field (~200 chars) to produce the recap — no need to read each article
  4. Group by source/topic when presenting results
  5. Only call feednest_get_article for articles the user specifically wants to read in full

Read an article (handle truncated RSS):

  1. feednest_get_article — get content
  2. If reading time is <1 min, content is likely truncated by the RSS feed
  3. Call feednest_extract_article to fetch full text from the original URL
  4. Do NOT extract every article — only truncated ones

Auto-tagging workflow:

  1. feednest_get_articles (unread, limit 20) — get today's articles
  2. For each article: analyze the title/content
  3. feednest_get_tags — check existing tags
  4. If needed: feednest_create_tag — create missing tags
  5. feednest_tag_article with action: "add" — apply tags

Highlight key passages:

  1. feednest_get_article — get full content
  2. Analyze and identify key passages
  3. feednest_add_highlight for each important passage (use exact article text, not paraphrased)

04 同仓库更多 Skills