Transcriptly gives video URLs an agent-ready transcript layer

Field notes

Transcriptly processes audio with Whisper rather than relying on scraped captions, then emits timestamped Markdown, SRT, or JSON. Its access points span a local npm-installed CLI, an MCP server, a hosted REST endpoint, and WebMCP tools registered directly on the page. The hosted service is limited to five transcriptions per hour and 30 minutes per video, while the page states that the CLI runs locally without those limits. The project is MIT licensed and links to its GitHub source.