Looking for an AI plugin for Obsidian means the community plugin browser, a search for "AI", and a lot of tabs. The directory can't filter by what a plugin actually does, and the names ("GPT", "Copilot", "Brain", "Companion") don't help.
So I made a list. Awesome-Obsidian-AI-Tools is a categorised index of Obsidian plugins that bring AI and LLMs into a vault. As of its last update on 15 December 2025 it lists 86 plugins in 17 categories, sorted by GitHub stars within each category, with a link to the source repository and an Obsidian install link for each.
GitHub Repository danielrosehill/Awesome-Obsidian-AI-Tools Obsidian plugins focused on leveraging AI & LLMs inside of (and beyond) vaults ★ 276Updated Dec 2025 View on GitHubWhat it is, and what it is not
It is a README. There is no code to install. Each entry has the plugin name, a star count, a link to its GitHub repo, a link into the Obsidian plugin directory, and the plugin's own one-line description.
It is also auto-organised, and I would rather say so plainly than let it pass as a review. The categorisation was done by a script that reads GitHub metadata (descriptions, topics, stars), with some manual overrides for repositories whose metadata was thin. I have not tested these 86 plugins. Some will be abandoned, some will be miscategorised, and a handful have no description at all (the README says "No description available" for those). Check the plugin's own page and recent commits before you install anything into a vault you care about.
What is in it
The categories are ordered by the combined stars of the plugins in them. The biggest are:
Autocomplete and writing assistance (2 plugins), which is where the most-starred entry, Copilot (about 5,800 stars at the time), lives.
Knowledge management and RAG (11 plugins), chat-with-your-notes and semantic search, including Smart Connections.
Local LLM integration (11 plugins), for Ollama and similar setups, including Local GPT and the Text Generator plugin.
Chat and conversation (20 plugins, the largest group by count).
The smaller categories are the more interesting ones if you have a specific need: canvas and visual workflows (6), multi-provider all-in-one plugins (6), speech and audio (6), learning and flashcards (4), summarisation (4), content generation (5), journaling (2), tagging and classification (1), graph visualisation (1), templates and prompts (2), and one under automation and computer control (Open Interpreter). There is also a small integration and utilities group, and a single provider-specific entry.
At the top of the README is a short stats block: 86 plugins, about 19,700 combined stars, and the five most popular by stars (Copilot, Smart Connections, Text Generator, ChatGPT MD and Local GPT).
How to use it
Open the README and scan the table of contents. Each category shows its plugin count and total stars.
Jump to the category that matches your need: local models, chat with notes, speech input, flashcards and so on.
Use stars and the last-updated date on each entry as a rough health check. A plugin with a recent update and some stars is a better bet than one with neither.
Follow the GitHub link to read the plugin's own documentation, then use the Obsidian link to install it.
If your plugin is not listed, the README also links the official directory searches for AI, GPT, LLM and Ollama.
Why a list and not a ranking
A recommendation goes stale quickly in a field this busy, whereas a categorised index is useful even when it is a little out of date, because it tells you what kinds of plugin exist. The useful question is usually "is there a plugin that does X, and one built on local models?" (voice notes into a vault, say), and the category headings answer that faster than the official directory does.
The list is also something I can regenerate. The repo's notes describe the process: pull fresh GitHub metadata with the GitHub API, run the categorisation script, rewrite the README. Adding a plugin by hand follows a fixed three-badge template, so entries stay consistent.
Limits
Nothing here is reviewed or tested. It is a map, not a verdict.
The data is a snapshot from December 2025. Star counts and update dates will have moved.
Categorisation by metadata is blunt. A plugin can sit in "Chat" when it is really a RAG tool, or the reverse.
It only covers plugins hosted on GitHub that have an entry in Obsidian's plugin directory.
If you spot a plugin that is missing or in the wrong place, the repository is public and the list is meant to be improved. The project is here: Awesome-Obsidian-AI-Tools.
