Show HN: I built a video transcript tool that also reads what is on screen

Category: ai-ml

Tags: mcp, video-analysis, transcription, ai-agents, multimodal

Score: 7.0/10 (Innovation: 6, Technical: 6, Documentation: 9, Utility: 7)

Scribiz is a hosted MCP server that gives AI agents structured access to video content—transcripts, summaries, and on-screen text—from YouTube, TikTok, Instagram, and other platforms. It stands out by combining multi-modal analysis (speech and visual) with timestamped citations, enabling agents to answer questions with precise video moments. The project is well-documented and integrates easily with popular MCP clients like Claude and Cursor.

Target audience: AI developers, backend engineers, and content analysts using MCP-enabled agents

Repository: https://scribiz.com · MIT · 1 stars

View on Hacker News