How therr-ctrl/video-scout-pipeline works
Video Scout Pipeline turns public Reddit stories and YouTube anecdotes into narrated videos, then uploads them privately for a manual review window. Its main workflow is scout → candidate queue → Gemini-assisted scripting → local rendering → quality-gated YouTube upload; the panel can also initiate work. The repository includes a separate canal2 planning and video-generation pipeline. Core workflow links are documented in the README; some canal2 render dependencies are supported by the source index, so unsampled behavior is not expanded.
Main entry points: pipeline.py Pipeline runner · generar_video_maestro.py Video renderer · script_writer.py Script writer · cola.py Candidate queue
Components 23
Story discovery
- Reddit scout
trend_scout.py - YouTube scout
youtube_scout.py
Queue and scripts
- Candidate queue
cola.py - Script writer
script_writer.py - Gemini client
gemini.py - Story scripts
script_writer.py - Long-story splitter
partir_historias.py - Content planner
canal2/content_planner.py - Canal2 script writer
canal2/script_writer.py
Video production
- Video renderer
generar_video_maestro.py - Narration
narrador.py - Background footage
hyperframes_broll.py - Rendered videos
generar_video_maestro.py - Canal2 renderer
canal2/generar_video_maestro.py
Review and operations
- Pipeline runner
pipeline.py - Local review panel
servidor.py - YouTube publisher
publisher.py - Canal2 publisher
canal2/publisher.py - JSON state
almacen.py
Other components
- Channel owner
- Reddit RSS
- YouTube sources
- YouTube
Connections 30
- Reddit RSS → Reddit scout: provides feeds
- Reddit scout → Candidate queue: adds candidates
trend_scout.py - YouTube sources → YouTube scout: provides captions
- YouTube scout → Candidate queue: adds candidates
youtube_scout.py - Candidate queue → Script writer: supplies candidates
script_writer.py - Script writer → Gemini client: rewrites stories
script_writer.py - Script writer → Story scripts: writes scripts
script_writer.py - Script writer → Long-story splitter: splits stories
script_writer.py - Story scripts → Video renderer: provides narration text
script_writer.py - Video renderer → Narration: generates speech
generar_video_maestro.py - Video renderer → Background footage: requests footage
generar_video_maestro.py - Video renderer → Rendered videos: writes videos
generar_video_maestro.py - Pipeline runner → Reddit scout: runs stage
pipeline.py - Pipeline runner → YouTube scout: runs stage
pipeline.py - Pipeline runner → Script writer: runs stage
pipeline.py - Pipeline runner → Video renderer: runs stage
pipeline.py - Pipeline runner → YouTube publisher: runs stage
pipeline.py - Rendered videos → YouTube publisher: supplies video
- YouTube publisher → Gemini client: checks metadata
publisher.py - YouTube publisher → YouTube: uploads privately
- Channel owner → Local review panel: reviews and controls
servidor.py - Local review panel → Video renderer: can start render
servidor.py - Candidate queue → JSON state: persists queue state
cola.py - Pipeline runner → Content planner: runs second pipeline
- Content planner → Canal2 script writer: plans then writes
- Canal2 script writer → Gemini client: uses model
canal2/script_writer.py - Canal2 script writer → Canal2 renderer: supplies planned scripts
- Canal2 renderer → Canal2 publisher: produces for publishing
- Canal2 renderer → Background footage: uses b-roll module
canal2/generar_video_maestro.py - Canal2 publisher → Gemini client: checks metadata
canal2/publisher.py