Daily AI Summary

Generate a daily arXiv digest using your own LLM settings.

What it does

• Fetches recent arXiv papers from your query
• Uses keyword prescreen + LLM judgment to decide what is interesting
• Writes a Markdown digest into a folder in your vault
• Can run manually, on startup, or on a fixed interval

Network and data use

• Calls the arXiv API to fetch recent papers and may download paper pages or source assets from arXiv to improve figure and reference extraction.
• Sends paper metadata and abstracts, your interest keywords, and matched historical paper summaries to the LLM endpoint you configure.
• Stores your configured LLM API key in the plugin data file inside your vault, as described in the settings UI.
• Writes generated digest notes, optional run logs, cached arXiv responses, and downloaded image assets into your vault.
• Does not include telemetry, ads, or any built-in third-party analytics.

External service requirements

• Requires network access to arXiv.
• Requires a user-provided LLM endpoint and API key.
• Works only on desktop Obsidian because the manifest sets isDesktopOnly to true.

Settings

• Target folder
• Summary title
• Interest keywords
• arXiv search query
• Max results
• Lookback hours
• LLM endpoint
• LLM API key
• LLM model
• LLM max completion tokens
• Run on startup
• Auto run interval minutes
• Ignore history

Development

npm install
npm run build

For watch mode:

npm run dev

Manual install

Copy these files into:

<your-vault>/.obsidian/plugins/daily-ai-summary/

Files:

• manifest.json
• main.js
• styles.css