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

# Integrations overview

> What your agents can do with each tool Leme connects to

Agents are only as useful as the tools they can reach. Each integration below gives your agents a new set of capabilities — reading your data, answering questions about it, and (where you allow it) taking action.

Every integration follows the same rules:

* **You connect once.** Sign in with the provider or paste a key; Leme stores the credential encrypted and never shows it to agents or other users. See [Connections](/connections).
* **You decide what agents may do.** Reading is allowed by default; anything that changes data or messages people asks a human first, until you decide otherwise. See [Permissions & approvals](/permissions).
* **Agents must be given access explicitly.** Connecting a tool does nothing until you enable it on an agent.

## Available integrations

<CardGroup cols={2}>
  <Card title="Slack" icon="slack" href="/integrations/slack">
    Read channels and threads, answer questions about discussions, post updates.
  </Card>

  <Card title="Gmail" icon="envelope" href="/integrations/gmail">
    Search and read your mailbox, draft replies, send email with your approval.
  </Card>

  <Card title="Outlook" icon="envelope-open" href="/integrations/outlook">
    Mail and calendar: triage the inbox, draft replies, schedule meetings.
  </Card>

  <Card title="GitHub" icon="github" href="/integrations/github">
    Browse repositories, issues, and pull requests; file issues and comment.
  </Card>

  <Card title="Linear" icon="list-check" href="/integrations/linear">
    Query, create, and update issues; keep projects moving.
  </Card>

  <Card title="Notion" icon="book" href="/integrations/notion">
    Search and read pages you share, query databases, write new content.
  </Card>

  <Card title="Google Analytics" icon="chart-line" href="/integrations/google-analytics">
    Traffic, conversion, and revenue reports from GA4, in plain language.
  </Card>

  <Card title="Database" icon="database" href="/integrations/database">
    Answer questions straight from your MySQL database — strictly read-only.
  </Card>

  <Card title="Axiom" icon="magnifying-glass-chart" href="/integrations/axiom">
    Explore logs and events: datasets, fields, and APL queries.
  </Card>

  <Card title="Stripe" icon="stripe" href="/integrations/stripe">
    Revenue, invoices, and customer questions with accurate totals.
  </Card>
</CardGroup>

<Note>
  Looking for Slack, Microsoft Teams, WhatsApp, or email as a place to *talk to* your agent? That's a [channel](/channels), not an integration. Integrations are tools agents work with; channels are where people and agents meet.
</Note>

## Shared and individual connections

Most integrations are connected once for the whole workspace — the team's Slack, the company's Stripe. Gmail and Outlook are different: they are **individual** connections. Each teammate connects their own mailbox, and an agent always acts through the mailbox of the person it is talking to — never someone else's. The full model is described in [Connections](/connections#shared-vs-individual).

## Read-only by design

Some integrations expose no write actions at all: **Google Analytics**, **Database**, **Axiom**, and **Stripe** are read-only today. Agents can answer any question the data supports but cannot change anything on the other side. For the rest, every action is classified by risk, and the connection's permission panel decides what runs freely, what asks a human, and what is off — see [Permissions & approvals](/permissions).