Instagram excels for quick, visual updates and Stories/Live from protests. Less ideal for long-form continuous feeds compared to Twitch or YT, but high engagement for rapid intel.
Current Limitations
- Heavy login/cookie requirements for reliable live access and scraping.
- No robust public embed API for persistent SPA integration (Reels are short-form).
- yt-dlp support exists but is fragile for lives; often needs fresh auth.
- Discovery is profile-based via Firecrawl + semantic agent searches rather than clean live directories.
- Ephemeral content; lives disappear quickly.
How to Import Video Feeds
- Discovery: Agent Firecrawl on Instagram profiles (e.g. bygeorgiafort) or x_keyword_search for "live" announcements. Use sources.json discovery_keywords.
- Capture: yt-dlp on profile/live URLs when active (cookies often required). Browser recording as fallback. Save to research/ for processing.
- Transcripts/Intel: Extract audio for Whisper; add visual descriptions or captions manually to research/tips/ or intel items. Run generate-data.py.
- Site: Add to active_protest_streams.json (platform: "instagram"). Display as linked source in Monitor. Full embed/player would need custom work (similar to YT persistent container).
- Cross-post detection: Many IG lives are announced on X; use the X research tools.
