Vai al contenuto
TakoAPI
Torna alle skill

veille

di romain-grosos

Web & Frontend Development
374Download riportati da clawskills.sh
2Stelle
0Visualizzazioni
0 likeAccedi per mettere like a questa skill
—Non ancora valutato
La tua valutazione

Sintesi

RSS feed aggregator, deduplication engine, and output dispatcher for OpenClaw agents.

Descrizione

Cosa fa questa skill

RSS feed aggregator that fetches articles from 20+ configured sources, removes already-seen URLs using a 14-day TTL store, and deduplicates stories covering the same topic via Jaccard similarity. Optionally scores articles with an OpenAI-compatible LLM and dispatches digests to Telegram, email, Nextcloud, or local files.

Requires no pip installs — pure Python stdlib — so it runs anywhere Python 3.9+ is available without environment setup.

Quando usarla

  • Morning security news digest from 20+ RSS feeds
  • Filtering duplicate stories about the same CVE or incident
  • Sending daily tech summaries to a Telegram channel
  • Archiving weekly digests to Nextcloud in append mode
  • Scoring and ranking articles by relevance with an LLM

Esempio di workflow

INPUT

User asks: 'donne-moi les news du jour'

AGENT
  1. 1

    Run veille fetch --filter-seen --filter-topic to pull new articles from configured RSS sources

  2. 2

    Pipe output through veille score to rank articles with the configured LLM

  3. 3

    Pass wrapped_listing to the LLM for a digest summary

  4. 4

    Run veille send to dispatch the digest to all enabled outputs

OUTPUT

Digest of new, deduplicated articles sent to Telegram, email, Nextcloud, or a local file

Requisiti

  • OpenAI-compatible API key stored in ~/.openclaw/secrets/ (required only if LLM scoring is enabled)
  • Telegram bot token (required only if Telegram output is enabled)
  • SMTP credentials (required only if using email output without the mail-client skill)
  • Nextcloud account and credentials (required only if using Nextcloud output without the nextcloud-files skill)

Installazione

clawhub install romain-grosos/veille

Esegui il comando CLI oppure incolla il prompt in Claude Code, Cursor, Codex o nell'agente di coding che usi.

Link