Alfred's Weekly Picks

A curated selection of thought-provoking essays, music, films, games, and more

Loading the collection...

Drop this file into your OpenClaw workspace or Claude Code project. Your agent will walk you through everything below — starting with a conversation about your interests and taste.

↓ Download CURATE.md

Or view it on GitHub

How It Works

Tell Your Bot What You Like

Give your Open Claw agent a brief on the kind of reading you want — topics, quality bar, tone, sources you trust. Be specific. The more you shape its taste, the better the curation.

Example prompt in Slack or your preferred channel:

"Find me 5 long-form articles this week. Topics: tech, culture, design, science — the weirder the better. Quality bar: New Yorker or pudding.cool level. Skip anything that reads like a press release."

Set Up a Sources List

Create a markdown file in your workspace with accounts and outlets you respect. Your bot will use these as seed sources to find articles — following them on X, checking RSS feeds, or browsing directly.

# sources.md ## Trusted Accounts (X/Twitter) - @robinsloan - @craigmod - @puloojafoundation ## Outlets - pudding.cool - theverge.com - longreads.com - aeon.co

Create Your Feed File

Ask your bot to generate a feed.json following the JSON Feed v1.1 spec. The key addition is the _picks extension — custom metadata that makes your curation machine-readable.

{ "version": "https://jsonfeed.org/version/1.1", "title": "Your Feed Name", "feed_url": "https://yourusername.github.io/picks/feed.json", "items": [ { "id": "unique-id", "url": "https://article-url.com/piece", "title": "Article Title", "summary": "Why this is worth reading.", "date_published": "2026-02-28T00:00:00Z", "tags": ["topic-a", "topic-b"], "_picks": { "selection_reason": "Natural language explanation.", "category": "reframe", "confidence": 0.9, "effort": "quick-read", "mood": "inspiring" } } ] }

The _picks fields are what let other bots and scripts understand why you chose something — not just what you chose. See the full extension schema for all available fields.

Host It on GitHub Pages

The simplest deploy: create a GitHub repo, drop in your feed.json and an index.html, and enable Pages. Your bot can do all of this for you.

# Ask your Open Claw bot: "Create a GitHub repo called 'my-picks', add the feed.json we made, build a simple reader page, enable GitHub Pages, and invite me as a collaborator."

Your feed is now live at https://yourusername.github.io/my-picks/feed.json — subscribable in any feed reader and consumable by any script.

Automate with a Cron Job

Set up an Open Claw cron job so your bot curates on a schedule — weekly, daily, whatever cadence you want. It'll find articles, generate the picks with metadata, update feed.json, and push to GitHub automatically.

# Example: weekly curation every Sunday at 10am "Set up a cron job: every Sunday at 10am ET, find 5 articles from my sources list, generate picks with _picks metadata, update feed.json, and push to GitHub."

Connect with Other Feeds

This is where it gets interesting. Once multiple people publish feeds with the _picks schema, anyone can aggregate them:

# A simple aggregator script could: 1. Fetch feed.json from 10 curators 2. Filter by category == "technical" and confidence > 0.8 3. Deduplicate by URL (overlap = strong signal) 4. Output a personalized reading list # No platform. No algorithm. Just URLs pointing to JSON.

Share your feed URL with friends, post it on your site, or register it in a community directory. The more feeds, the richer the network.

Share Your Feed (Directory Optional)

Your feed URL is all anyone needs to subscribe — share it however you want.

Test it in the reader: Visit the Picks Reader, paste your feed URL, and see how it renders alongside other curators' picks.

Optionally, submit to the directory: The Feed Directory makes it easier for people to discover new curators. Submit via GitHub issue. Note: the directory is a convenience, not a requirement — it's technically a centralization point. Your feed works perfectly fine without it.

The schema, source code, and a starter template are all in the GitHub repo. Fork it and make it yours.

Start Your Own Feed