Hey PH 👋 We're the team behind ViralMint — sharing our first big public launch today. We built this because we kept stitching seven different tools every week to publish faceless shorts — vidIQ for trends, yt-dlp for downloads, Opus for clips, ElevenLabs for voice, CapCut for captions, Pictory for stock assembly, separate copy-paste for titles and tags. Every step had its own subscription and its own export friction. ViralMint runs the whole pipeline from one desktop app: scout trending videos across YouTube + TikTok + Douyin → analyze competitors with local Whisper → generate original 9:16 mp4s with AI script, voice, word-by-word captions, and AI-generated or stock clips. No watermark, no per-minute cap, pay only for what you generate (~$0.30–$1 per finished short). What we shipped this week: a 67-tool MCP server. Hook up Claude Code, Cursor or Claude Desktop and you can drive the entire pipeline conversationally — "scout AI tools, download the top 3, generate a 60s Short from the winning hook" and Claude chains scout → download → extract clips → generate → export automatically. Cost preview before spending, real-time activity feed, user-defined slash-prompts. Open-source at github.com/openclaw-easy/ViralMint under AGPL-3.0 — the desktop is fully auditable, BYOK supported (OpenAI / Anthropic / OpenRouter). The cloud is the optional hosted-convenience layer; you can skip it entirely. Would love your honest take — especially if you're a creator stuck in the multi-tool stitch we described, or if you've tried hooking MCP up to other workflows. We're around all day to answer anything about the architecture, the open-core split, or the per-tier video-model pricing. -Toni

ViralMint

Open-source AI video pipeline. Claude Code drives it

Open-source AI video pipeline. Claude Code drives it