---
title: "Trove documentation"
description: "Use when you are new to Trove and need to know what it is, how it works in three steps, and which page to open next."
url: https://docs.heytrove.ai/
updated: 2026-11-12
---

# Trove documentation

> Load this page when: you need a first overview of Trove or do not yet know which docs page to load

## For agents

- Reach Trove through the `trove` CLI (coding agents) or the MCP server (Claude and Grok on the web). ChatGPT installs Trove from its app directory.
- Start with `trove items list --json --no-session` to see what the library holds.
- Load one page, not the whole site. The index is `https://docs.heytrove.ai/llms.txt` and the full dump is `https://docs.heytrove.ai/llms-full.txt`.

Trove is a [library](https://docs.heytrove.ai/concepts/library-and-books.md) your AI agents read from. You add books, docs and notes, and your agent finds the right pages, reads them, and answers with [citations](https://docs.heytrove.ai/concepts/citations.md) you can check.

## How it works

1. You ask your agent to add knowledge: PDF, EPUB or Markdown files, or books from the [marketplace](https://docs.heytrove.ai/concepts/marketplace.md).
2. The [librarian](https://docs.heytrove.ai/subagents/librarian.md) picks the books that fit your task and returns a reading list.
3. [Item readers](https://docs.heytrove.ai/subagents/item-reader.md) open those pages and answer with the book title and page numbers.

The most reliable way to get knowledge from Trove today is to ask for it. Tell your agent to search your Trove library, for example: "Search my Trove library for how we handle retries." That starts the librarian. The librarian picks the books and pages, and it starts the readers on its own. You do not start readers yourself. Trove also has automatic checks, and we are working to make the librarian start on its own. For now, do not rely on them. When the answer matters, ask.

Want your agent to follow Trove best practices? Ask it: "Read How to Use Trove in my Trove library and follow it." That book is in every library.

## Install

Run one command in your terminal. The [installer](https://docs.heytrove.ai/install.md) sets up the `trove` command and signs you in through the browser. You do this one time. After that, you tell your agent what to do, and your agent runs Trove.

```terminal
curl -fsSL https://heytrove.ai/install.sh | sh
```

On Windows, run `powershell -c "irm https://heytrove.ai/install.ps1 | iex"` in PowerShell. Step-by-step instructions for every agent are in [Install in your agent](https://docs.heytrove.ai/install.md).

## Where Trove runs

| Where | How it connects |
| --- | --- |
| Claude Code, Codex, Cursor, GitHub Copilot (CLI and VS Code), Gemini CLI, Grok | The `trove` CLI and a skill that the installer adds to each agent |
| Claude (app, web, desktop, Cowork) and Grok on the web | The MCP server at `https://heytrove.ai/api/v1/mcp` |
| ChatGPT | The Trove app in ChatGPT's app directory, see [Install in your agent](https://docs.heytrove.ai/install.md) |

## Where to next

- [Quickstart](https://docs.heytrove.ai/quickstart.md): add one book and ask your first cited question.
- [Set up your library](https://docs.heytrove.ai/usage/set-up-your-library.md): turn the skills you already have into books, then add marketplace books.
- [Best practices](https://docs.heytrove.ai/best-practices.md): what to add, how to organize it, and how to keep answers trustworthy.
- [Usage](https://docs.heytrove.ai/usage.md): step-by-step guides for common jobs.
- [Concepts](https://docs.heytrove.ai/concepts.md): the nine ideas behind Trove, one page each.
- [Subagents](https://docs.heytrove.ai/subagents.md): who does what when your agent researches.
- [Reference](https://docs.heytrove.ai/reference.md): every command, API route and limit.
