Grand Aim

A semantically colored feed of bluesky skeets, for your perspective, someone else's or another group's.

Intermediate goals so far

• Visualize bluesky data.
• Read from jetstream
• Extract interesting fields
• Filter by commit/post type
• Get color selection ergonomic for further development, 24-bit? 32-bit? color, the alpha paramter of colorama is quite useful
  • colorspace, hsla / hsl
  • Vary colors, find balance so nothing draws excessive attention, but noticeable, though gradual, color evolution occurs And, for ergonomic use to communicate meaning. Color for cluster / direction. Brightness, inverse to length, for ease of reading, uniformity of gestalt. Saturation for content searching versus scaffolding metadata. Errors in grey too.
• Get semantic clustering working -- static color. Faiss? enough? Word2Vec? Queries?
• Get sematic selection & filtering working, based on keywords, chosen posts
• Get db/profiles/queries/modes set up -- so every taste selection input can inform any future use
• Get a streaming web view working -- possibly even fully in-browser
  • fallen? twitterfall inspiration. hatnote's L2W

Run it

uv run --with websockets,colorama,munch,fire python3 bluesky-simple-print.py

What it looks like