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

# Integrations Overview: Apps, Skills, Credentials & Browsers

> Learn how Kolmena connects your agents to apps, skills, credentials, and browsers so they can act on your behalf using real data.

Kolmena gives your agents the ability to work with real-world services through integrations. You control what each agent can reach, what accounts it signs into, and what actions it may take.

The platform groups these capabilities into four areas:

* **Apps (MCP servers)** — external services such as Gmail, Slack, Notion, GitHub, and many more. Agents call these through a catalog of integrations you enable per workspace or per agent.
* **Skills** — reusable capabilities you can install, write, or upload. Skills extend what an agent knows how to do, from web scraping to custom business logic.
* **Credentials** — API keys and tokens your agents use to sign into services. Stored encrypted, they are never shown again after you save them.
* **Browsers** — a paired browser extension that lets agents work inside live web pages you are already signed into.

<CardGroup cols={2}>
  <Card title="Connect Apps" color="#FE4F32" icon="plug" href="/integrations/connect-apps">
    Browse the catalog, connect an app with OAuth or an API key, and enable it for Maya or any agent.
  </Card>

  <Card title="Skills" color="#FE4F32" icon="wrench" href="/integrations/skills">
    Browse, install, and manage skills that extend what your agents can do.
  </Card>

  <Card title="Credentials" color="#FE4F32" icon="key" href="/integrations/credentials">
    Add credentials, grant them to agents, and handle credential requests from chat.
  </Card>

  <Card title="Browsers" color="#FE4F32" icon="globe" href="/integrations/browsers">
    Pair a browser so agents can use live tabs and sites you are already signed into.
  </Card>
</CardGroup>

## How agents use integrations

Agents do not have blanket access to everything. You decide:

1. **Which integrations are available** — enable an app or skill for all agents, or attach it to one agent only.
2. **Which credentials are granted** — nothing is granted by default. You explicitly choose which agents may use each credential.
3. **Which actions are allowed** — for every app, you can narrow the list of tools the agent may call.

Changes apply immediately. A running agent picks them up on its next reply, so there is no publish step for integrations.

## Quick comparison

| Area | What it is | Where you manage it |
| - | - | - |
| Apps | External services (Gmail, Slack, etc.) | **Integrations** page, or per agent under **Configure** |
| Skills | Reusable capabilities and custom logic | **Skills** page, or per agent under **Configure** |
| Credentials | Encrypted API keys and tokens | **Settings > Credentials** |
| Browsers | Live browser tabs for agent use | **Settings > Your browsers** |


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.