> ## Documentation Index
> Fetch the complete documentation index at: https://wiki.mira.tg/llms.txt
> Use this file to discover all available pages before exploring further.

# Agentic Infrastructure

> Mira is built to act, not just respond

## Overview

Most AI tools answer questions. Mira executes tasks. The difference is infrastructure — and Mira's is built from the ground up for autonomous action inside Telegram.

## Skills

With **Skills**, users automate work and daily routines — reminders, data pulls, recurring reports — and ask Mira to proactively run them on a schedule.

Skills run in the most natural place for agent communication: **Telegram**. No dashboards, no cron jobs, no code.

```
Every Monday at 9am, pull my GitHub issues and summarize them
Remind me to review the deck 1 hour before every meeting
Check my inbox daily and flag anything urgent
```

## Integrations

Through **APIs and MCP integrations**, Mira connects to:

<CardGroup cols={3}>
  <Card title="Wallets" icon="wallet">
    TON, USDT, on-chain actions
  </Card>

  <Card title="Notion" icon="book">
    Pages, databases, docs
  </Card>

  <Card title="Calendar" icon="calendar">
    Events, scheduling, reminders
  </Card>

  <Card title="Gmail" icon="envelope">
    Read, send, search emails
  </Card>

  <Card title="GitHub" icon="github">
    Repos, issues, pull requests
  </Card>

  <Card title="500+ more" icon="plug">
    Via Composio integration layer
  </Card>
</CardGroup>

Mira orchestrates **tools, memory, and actions** inside Telegram — so users never have to leave the chat.

## Interchangeable LLMs

Mira is model-agnostic. LLMs are interchangeable components:

| Model                | Provider             |
| -------------------- | -------------------- |
| Claude Opus / Sonnet | Anthropic            |
| GPT-4o               | OpenAI               |
| Qwen3-32B            | Alibaba (via Cocoon) |
| Gemini               | Google               |

The right model is selected based on the task, user preference, or privacy requirements.

## Agent-to-Agent

Users can connect **specialized agents** to Mira for professional tasks and multi-step automations — chaining with external LLMs or domain-specific agents for complex workflows.

<Note>
  Agent-to-agent interaction enables Mira to delegate subtasks to specialized tools and return a unified result to the user.
</Note>