noscroll 0.1.2

Pull, don't scroll. RSS aggregator with LLM-powered summarization.

NoScroll - Pull, don't scroll

What is NoScroll

NoScroll is a Python CLI that pulls information from RSS feeds, web pages, and Hacker News, then uses an LLM to summarize and rank the most useful items.

It is designed for a pull-based reading workflow: define sources once, run on schedule, read only the high-signal digest.

Installation

pipx install noscroll

Ask Command

Use natural language directly:

noscroll --env-file .env ask "Collect content from the past five days, one file per day"

This will generate daily digest files in outputs/.

Example generated text:

## AI (3)
1) Off Grid: Running text/image/vision models offline on mobile | Value: 4/5 | Type: Practice
- Conclusion: This open-source project demonstrates on-device multimodal inference on smartphones, with strong privacy and offline usability.
- Why it matters: On-device AI can reduce privacy risk and cloud inference cost, and is a good fit for offline-first products.
- Evidence links: https://github.com/alichherawalla/off-grid-mobile

## Other News (2)
4) uBlock rule: hide YouTube Shorts with one click | Value: 4/5 | Domain: Tech

## Life & Health (2)
6) AI avatars for rural healthcare support | Value: 3/5 | Domain: Health

Configuration

You can provide a config file in these places:

1. CLI argument: --config /path/to/config.toml
2. Environment variable: NOSCROLL_CONFIG=/path/to/config.toml
3. Default path: ~/.noscroll/config.toml

Example config.toml:

[llm]
api_url = "https://api.openai.com/v1"
api_key = "your-api-key"
model = "gpt-4o-mini"

[paths]
subscriptions = "subscriptions/subscriptions.toml"
output_dir = "outputs"

[runtime]
debug = false

Create a starter config:

noscroll init

License

MIT. See LICENSE.