Text-Gen

Text-Gen is an AI tool. Open-source Obsidian plugin that adds AI text generation to your notes through reusable templates, with support for GPT-4 Turbo, Claude, Bard and Llama.

Text-Gen is an open-source plugin that brings generative AI into Obsidian, the markdown-based note-taking app. Instead of switching to a browser tab to query a chatbot, the plugin lets users generate, expand and reshape text inside their own vault, where notes, links and templates already live. The project is aimed at people who treat Obsidian as a personal knowledge base: researchers, writers, students and anyone who wants AI assistance without exporting their thinking to a separate tool. The mechanism is template driven. Users write prompts as templates and attach metadata that injects context, so the same template can produce different output depending on the note it runs in. A shared template hub lets users exchange these templates with other people, which turns prompt design into something reusable rather than a one-off chat message. The plugin supports a range of language models, including GPT-4 Turbo with a 128k context window, GPT-3.5 Instruct, Claude, Bard and Llama, so users can pick a backend that matches their budget or privacy needs. In practice, Text-Gen suits recurring writing tasks: summarizing a long note, drafting sections from an outline, or generating variations of a paragraph. Because it runs inside Obsidian, the output lands in the file system as plain markdown rather than being trapped in a chat history. The trade-off is that it is a plugin, not a standalone product: it requires an Obsidian install, a configured model provider and some patience with template syntax. Users who want a polished chat interface or built-in model access without setup will find dedicated writing assistants simpler, while Obsidian power users get a tighter loop between notes and generation. Compared with general AI writing tools, Text-Gen competes on integration and openness rather than model quality. It is free, the code is public, and community templates shorten the learning curve.