infer audio repo during upload

fallback synthesize on edge tts failure

fallback script on zerogpu expiry

apply audio repo prefs to backend

add history latest podcast download

add latest podcast download link

avoid closing duckdb during bootstrap

extend zerogpu script duration

preserve local duckdb on missing remote

harden duckdb sync path creation

upload initial duckdb for existing dataset

fix duckdb initial sync file

fix duckdb initial sync file

fix stage auth request forwarding

fix frontend auth status bootstrap

fix frontend auth status bootstrap

fix frontend auth status bootstrap

fix(zerogpu): normalize tiny-model script speakers

fix(zerogpu): persist raw rank diagnostics and script indexes

fix(zerogpu): keep fallback rank items above threshold

fix(web): pass resolved run id into background pipeline

fix(rank): keep current-run items when dedup overfilters

test(config): shrink zerogpu podcast generation inputs

fix(api): allow hf token fallback for stage tests

fix(space): import zerogpu gpu functions at startup

fix(space): keep gradio process alive

fix(space): use canonical gradio demo entrypoint

fix(space): enable gradio queue for zerogpu

fix(space): pin gradio 5 for zerogpu runtime

fix(space): expose gradio demo for zerogpu

test: add zerogpu metrics runner for small models

refactor: unify storage layer and modularize Gradio web app

feat: ZeroGPU backend for Hugging Face Spaces deployment (#4)

fix: launch pipeline via Hugging Face Jobs

fix: hydrate Space DuckDB after OAuth login

fix: persist stage diagnostics and accurate run status

fix: request inference scope for Space OAuth

fix: pass Jina API key to reader requests

fix: propagate session token to HF clients

fix: use Space OAuth token for pipeline API

fix: use Hugging Face inference on Space

feat: add scheduled job backend and resumable pipeline sync

chore: hardening

fix: 8 critical bugs, config hardening, pipeline structural fixes, dead code cleanup (#3)

chore: simplify stage_fn to use run_single_stage directly

feat: implement background task worker queue and pollable history view to run pipeline asynchronously

feat: split pipeline trigger UI into two separate buttons to issue a fresh ZeroGPU proxy token before script writing

feat: implement automatic CPU fallback for ZeroGPU queue lease execution failures

feat: implement frontend pipeline stage chaining to prevent ZeroGPU request lease timeouts

Optimize ZeroGPUBackend model loading by caching weights in system CPU RAM at startup/request start, moving to GPU instantly only during leases to bypass Expired ZeroGPU proxy token timeouts