{"id":47217763,"url":"https://github.com/lucidprogrammer/youtube-vision-transcriber","last_synced_at":"2026-03-13T16:35:37.648Z","repository":{"id":333464744,"uuid":"1136615059","full_name":"lucidprogrammer/youtube-vision-transcriber","owner":"lucidprogrammer","description":"AI-powered pipeline that converts YouTube videos into polished articles using vision-based transcription - captures code, terminal output, and on-screen text that subtitles miss","archived":false,"fork":false,"pushed_at":"2026-01-19T16:57:03.000Z","size":254,"stargazers_count":0,"open_issues_count":1,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-01-19T18:13:30.393Z","etag":null,"topics":["ai","fast-agent","gemini","knowledge-","knowledge-extraction","llm","mcp","model-context-protocol","openai","python","transcription","video-to-text","vision-ai","youtube"],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/lucidprogrammer.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-01-18T02:26:56.000Z","updated_at":"2026-01-19T16:57:08.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/lucidprogrammer/youtube-vision-transcriber","commit_stats":null,"previous_names":["lucidprogrammer/youtube-vision-transcriber"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/lucidprogrammer/youtube-vision-transcriber","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lucidprogrammer%2Fyoutube-vision-transcriber","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lucidprogrammer%2Fyoutube-vision-transcriber/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lucidprogrammer%2Fyoutube-vision-transcriber/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lucidprogrammer%2Fyoutube-vision-transcriber/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/lucidprogrammer","download_url":"https://codeload.github.com/lucidprogrammer/youtube-vision-transcriber/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/lucidprogrammer%2Fyoutube-vision-transcriber/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":30471103,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-03-13T11:00:43.441Z","status":"ssl_error","status_checked_at":"2026-03-13T11:00:23.173Z","response_time":60,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.6:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai","fast-agent","gemini","knowledge-","knowledge-extraction","llm","mcp","model-context-protocol","openai","python","transcription","video-to-text","vision-ai","youtube"],"created_at":"2026-03-13T16:35:36.404Z","updated_at":"2026-03-13T16:35:37.640Z","avatar_url":"https://github.com/lucidprogrammer.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# YouTube Vision Transcriber\n\n**AI-powered pipeline that converts YouTube videos into polished articles using vision-based transcription.**\n\nUnlike standard transcribers that only read audio captions, this tool uses **Google Gemini 2.0 Flash** (via `fast-agent`) to \"watch\" the video, capturing code snippets, terminal output, and on-screen text that subtitles often miss.\n\n## Features\n\n- **Vision-Based Transcription**: Captures visual context, code blocks, and diagrams.\n- **MCP-Native**: Built with the [Model Context Protocol](https://modelcontextprotocol.io), compatible with Claude Desktop and VS Code.\n- **Robust Pipeline**:\n    - **Video Preparer**: Downloads and splits videos into manageable chunks.\n    - **Part Transcriber**: Transcribes each chunk with timestamped visual details.\n    - **Doc Writer**: Synthesizes a structured technical article from the transcript.\n- **Fast \u0026 Efficient**: Uses `uv` for dependency management and `fast-agent` for orchestration.\n\n## Prerequisites\n\n- **Python**: \u003e= 3.13.5\n- **uv**: [Install uv](https://docs.astral.sh/uv/getting-started/installation/)\n- **API Keys**:\n    - **Google Gemini**: For vision transcription.\n    - **OpenAI** (Optional): For orchestration (default model is `gpt-4o-mini`, but can be changed).\n\n## Installation\n\n1.  **Clone the repository**:\n    ```bash\n    git clone https://github.com/lucidprogrammer/youtube-vision-transcriber.git\n    cd youtube-vision-transcriber\n    ```\n\n2.  **Install dependencies**:\n    ```bash\n    uv sync\n    ```\n\n3.  **Configure API Keys**:\n    Copy the example secrets file and edit it:\n    ```bash\n    cp fastagent.secrets.yaml.example fastagent.secrets.yaml\n    ```\n    Add your API keys to `fastagent.secrets.yaml`:\n    ```yaml\n    google:\n      api_key: \"YOUR_GEMINI_API_KEY\"\n    openai:\n      api_key: \"YOUR_OPENAI_API_KEY\"\n    ```\n\n4.  **Install MCP Sub-Server Dependencies** (For Local Development):\n    The project uses external MCP servers that need to be installed globally:\n    ```bash\n    # Filesystem MCP server (Node.js)\n    npm install -g @modelcontextprotocol/server-filesystem\n    \n    # Fetch MCP server (Python)\n    pip install mcp-server-fetch\n    # OR using uv:\n    uv pip install mcp-server-fetch\n    ```\n    \n    \u003e **Note**: If you're only using Docker, you can skip this step as these dependencies are pre-installed in the container.\n\n## CLI Usage\n\nYou can run the agent interactively in your terminal:\n\n```bash\nuv run agent.py\n```\n\nPaste a YouTube URL when prompted. The agent will:\n1.  Download the video to `./youtube_data`.\n2.  Split and transcribe it.\n3.  Generate an `article.md` in the video's folder.\n\n### 🚀 Quick Start (Docker)\n\n1. **Pre-create a Persistent Volume** (Recommended for data persistence):\n   ```bash\n   # Create a Docker volume linked to your home folder for easy results access\n   mkdir -p ~/youtube_data\n   docker volume rm youtube_vision_vol || true\n   docker volume create --driver local \\\n     --opt type=none \\\n     --opt device=$HOME/youtube_data \\\n     --opt o=bind \\\n     youtube_vision_vol\n     \n   ```\n\n2. **Configure MCP Client**:\n   Add this to your `windsurf.json`, `cursor.json`, or `claude_desktop_config.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"youtube-vision\": {\n      \"command\": \"docker\",\n      \"args\": [\n        \"run\",\n        \"-i\",\n        \"--rm\",\n        \"-e\", \"GOOGLE__API_KEY=your_key_here\",\n        \"-e\", \"OPENAI__API_KEY=your_key_here\",\n        \"-v\", \"youtube_vision_vol:/app/youtube_data\",\n        \"lucidprogrammer/youtube-vision-transcriber\"\n      ]\n    }\n  }\n}\n```\n\n### 🛠️ Troubleshooting (Docker Mounts)\n\nIf you see a **\"mounts denied\"** error (e.g., `The path /tmp/youtube_data is not shared`):\n\n1. **Host Sharing**: In Docker Desktop settings, go to **Resources \u003e File Sharing** and ensure the path is added.\n2. **Path Selection**: Use a path inside your home directory (e.g., `/Users/yourname/youtube_data`) which is usually shared by default.\n3. **Absolute Paths**: Ensure you use the full absolute path on the host machine.\n\n## Architecture\n\nThe system consists of three main MCP components orchestrated by a master agent:\n\n1.  **YouTubeVisionTranscriber (`youtube_mcp.py`)**: Handles downloading (`yt-dlp`) and splitting videos (`ffmpeg`).\n2.  **VideoTranscriberServer (`video_transcriber_mcp.py`)**: Wraps internal LLM calls to process video chunks.\n3.  **Filessytem Server**: Manages reading/writing data to `./youtube_data`.\n\n## License\n\nMIT License. See [LICENSE](LICENSE) for details.\n\n## Contributing\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Flucidprogrammer%2Fyoutube-vision-transcriber","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Flucidprogrammer%2Fyoutube-vision-transcriber","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Flucidprogrammer%2Fyoutube-vision-transcriber/lists"}