Fusion AI, powered by HAL 8

Browse Source

The hub's assistant becomes a consumer HAL 8 chat:
- The button is HAL 8's red eye (the real logo); the header reads POWERED BY HAL 8.
- One control, a Standard | Incognito pill. Standard keeps the conversation in this browser;
  Incognito keeps nothing and asks the server for no retention.
- Always on, no toggles: chat and code (copy; Python runs in the browser via Pyodide), images and
  video inline, web answers with their sources, voice in (speech recognition) and read-aloud.
- Streams from HAL 8's public Fusion endpoint (hal8.odyssey-works.io/v1/public/fusion/chat); the
  browser never holds a key. Limits come back as a friendly "see you tomorrow" card.
- A small safe markdown renderer: everything escaped, https-only links and media, HAL 8-relative
  file links resolved against HAL 8.
- tools/fusion_inject.py replaces the old assistant block idempotently (between /*F8-BEGIN*/ and
  /*F8-END*/) and inlines fusion/fusion.css.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

...

This commit is contained in:

jerry

2026-09-26 18:50:44 -05:00

parent f50457b1c1

commit 716615a1d2

4 changed files with 423 additions and 1 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

2

index.html

File diff suppressed because one or more lines are too long