---
title: "Research Library"
id: "1737"
type: "page"
slug: "researches"
published_at: "2026-09-20T20:56:05+00:00"
modified_at: "2026-09-21T01:36:28+00:00"
url: "https://xedant.com/agents/research/docs/researches"
markdown_url: "https://xedant.com/agents/research/docs/researches.md"
excerpt: "The Research Library is the analyst’s workspace: a tree of Markdown documents organized by category,…"
---

# Research Library

[https://xedant.com/agents/research/docs/researches.md](https://xedant.com/agents/research/docs/researches.md)

The Research Library is the analyst’s workspace: a tree of Markdown documents organized by category, a built-in editor, and the source materials behind every report. Research results, plans, and cycle documents all land here.

## Document tree

Documents are organized into a tree of categories with navigation. Every file carries a frontmatter header with title, order, and description: the tree builds its labels and ordering from these. The tree is re-read whenever something changes, so new reports and edits appear instantly — updates happen on the fly, with no reload.

## History, search, and filters

In a large library, documents stay easy to find. Next to the search box in the sidebar sits a clock button: it switches what the list shows — the full tree, recently viewed, recently updated, or recently created documents. “Recently Viewed” is stored on this device, while “Updated” and “Created” are derived from the files’ timestamps.

Search is full-text: as you type, it looks through titles, paths, and document bodies. You can also filter the library by document properties (frontmatter) — for example, keep only the documents tagged with a certain tag. You compose such queries in the advanced search dialog, and the available field values are suggested for you automatically.

Any document can be given an icon from the built-in collection — the category tree becomes easier to read, and the section you need is found at first glance. Each page remembers its scroll position for a day: open a document, and you are exactly where you left off. At the bottom of the sidebar are the “Prompts (N)” and “Chats” buttons: the first opens the shared prompts queue (see “[Prompt Queue](/agents/research/docs/prompts-queue)
“), the second the list of agent chats.

## Editor

The built-in editor with syntax highlighting (CodeMirror) lets you edit any document right in the browser — reports, plans, cycles. It is plain Markdown: files can be edited, deleted, and extended without restrictions.

## Report sources

Every report shows its history: the search sessions and the downloaded pages it was assembled from. The downloaded Markdown is read right in the interface — you can check any fact against the original in one click, without leaving the app.

## Edit with AI and “Call Agent”

Next to the edit button sits an AI icon. It opens a panel with a prompt field: describe what to change in the document, and the prompt goes to Xedant Agent (a request to `{AGENT_API_URL}/api/remote/chats`), which performs the edit. The absolute path of the research is substituted into the prompt automatically, and the model is picked from the Xedant Agent’s model list, which the Research Agent reads live over the API. The finished agent chat opens from the interface.

The feature is enabled by the `AGENT_API_URL` and `AGENT_API_KEY` variables (set together); without them the buttons are hidden. Learn more in [Xedant Agent Integration](/agents/research/docs/agent)
.

## Agent chats

The “Chats” panel complements the library: Xedant Agent chats with real-time updates — the agent works on the documents while you see every step in the same window.

[← Back to documentation](/agents/research/docs)