Point it at a video, image, or PDF — get structured JSON. Local or cloud.
Vidlizer is a CLI + MCP server that extracts structured JSON from any visual media. Feed it a video, image, or PDF and get a machine-readable user journey, event timeline, or content analysis back. Runs fully local via…
Design decisions
- → Local-first, cloud-optional: Ollama and LM Studio support means zero API costs for private video analysis. OpenRouter provides cloud fallback for teams without local GPU capacity.
- → MCP server for IDE integration: Runs as an MCP server inside Claude Code, Cursor, and Claude Desktop. Agents call it as a tool — no copy-paste, no manual file handling.
- → Apple Silicon optimized: MLX Whisper for transcription on M-series chips. Frame extraction and model inference tuned for macOS Metal. Significantly faster than cloud round-trips for local workflows.