Release v1.4.0 · T04435/ghostq

GhostQ v1.4.0, live at https://ghostq.xyz. CollabQ: one song for two crowds.

What's new

CollabQ ("A feat. B"). Under Artist there's a new, optional feat. field (the same Spotify search). With a second artist, GhostQ reads both crowds from Qloo, then asks Qloo what the two crowds share.
A verse per crowd, a chorus for both. Verse 1 is written for A's crowd in A's voice, verse 2 for B's crowd in B's voice, and the chorus is built on what the crowds share, sung by both.
Bilingual when the crowds are. Each verse is in its artist's language; when the languages differ, the chorus mixes both.
Whose signal is whose. Every Qloo receipt shows the crowd it came from: A, B or Shared. The page shows both crowd cards.
Labelled on every screen. The song page, Library, share preview and Markdown export all read "GhostQ draft for A feat. B, AI-generated, not affiliated with A or B". Singing is blocked if the lyrics name either artist.
A tidier research trace. Repeated calls to the same Qloo tool fold into one step (for example "Find the artists on Qloo · 2 calls"). Expand it to see each call.

/search and /v2/insights (heatmap, demographics, tags, artists, films, books, audiences) run for each artist, as for a solo song.
/v2/analysis/compare, with one artist on each side and filter.type=urn:tag, returns the tags both crowds share, ranked by score, plus each side's own tags. The chorus is written from the shared ones.

A CollabQ run makes about twice the Qloo calls of a solo song, so it counts twice against the daily research budget.
When Qloo has no data on one of the artists, GhostQ asks for it the same way, once per artist.

Setup and keys: README. What's next: ROADMAP.md.

Licence: MIT.