This is the complete guide to Xedant Agent — a short note on what each section does and why you need it. If you have just installed the app, start with Getting Started.
Getting Started
Where to start: installation, first login, a tour of the interface, and your first message.
Validation
How the agent checks its own work: chat output, Git changes, builds, deployment, hooks, analytics and more.
AutoFix
Automatic error discovery and fixing: how to enable it, how to watch it work, and what its limits are.
Git
Working with version control: tracking changes, committing, branches, and pushing to a remote.
Build & Deploy
Building and publishing the project: the build panel, auto-builds, deployment, and reading errors.
Skills
A skill is attached to the chat and loads automatically: picking one in the input bar, managing the catalog, export and import.
Chat
Sending messages, understanding replies, choosing skills, and the message queue.
Files
The file tree, the code editor, and creating, renaming, moving and deleting files.
Models
Connecting AI models from different providers, switching between them, the peak-hours indicator, and auto-send rules.
Hooks
The rules the agent follows: what is allowed and forbidden, automatic commands, and tool control.
MCP Servers
Connecting the agent to outside services and databases through the Model Context Protocol.
Prompts
Saved requests with automatic sending: the prompt queue, AutoSend, and settings.
Context Utilization
How much text the model remembers, how to watch the usage, and how to keep long conversations from losing quality.
Commits
The history of saved changes across branches, comparing file versions, and commits linked to their chats.
Scripts
Running ready-made scripts, with live output and the ability to stop at any moment.
Mobile
Agent is a web app first: a responsive interface plus installation on your phone or desktop.
Remote Control
Accessing the agent from other programs: creating chats, sending tasks, and embedding AI into your services.
Analytics
Usage charts, metrics, date ranges, code statistics, and chat time per skill.
Telegram Bot
Controlling chats from Telegram: inline buttons, tool progress, and reasoning.
Settings
The settings card hub: appearance, chat, automation, tools, configuration, MiniAgent profiles, and account.
Shortcuts
The complete keyboard shortcut reference for power users.
.xedant Folder
Where project settings live: config files, build and deploy targets, parsers, and service data.
Licensing
License types, activation, status information, and updating your license key.
Proxy Server
The built-in proxy for traffic logging, protocol conversion, token counting, and cost calculation.
Tips & Troubleshooting
Productivity tips, common questions, fixing problems, and getting help.
Simplified Interface
The minimalist ChatGPT-like mode: message bubbles, hidden tools, and simplified input.
Backends
Built-in engines: Claude Code, Codex, OpenCode, Pi, Oh My Pi, Reasonix, DeepSeek Harness, Kimi Code, the built-in MiniAgent, and Command Code — switching through the model picker, auto-installation, and adding new ones.
SSL Certificates
Free Let’s Encrypt certificates issued right from the interface, plus self-signed certificates for working by IP without a domain.
Tasks
Scheduled automations: scripts, launch conditions, AI analysis of results, and delivery through a bot.
Bots
Your own Telegram bots: each with its own model, skill, system prompt, and allow-list of users.
API
Programmatic access to chats and the queue: key authentication, /api/remote commands, the extended surface, and live SignalR events.
MiniAgent — the built-in agent
Fully self-sufficient out of the box: built into Xedant Agent, nothing to install, no dependence on third-party developers. Tools, sessions, skills, and context are available right away.
Voice Input
Dictating messages by voice: the browser engine, local Whisper, cloud services, and your own OpenAI-compatible services.
Chat Labels
Colored labels on chats: an icon before the name, list filtering, built-in and custom labels, and automatic question and error labels.
Login & Authorization
Ways to access the agent: regular login, single sign-on through your server (SSO), API-key authorization, and live connections.