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

# Built-in Capabilities Overview

> Discover all 8 built-in capabilities available in Noorle Platform. Learn what each does and how to enable them.

Noorle Platform includes 8 built-in capabilities that provide core functionality without requiring external integrations. Attach them to your agents and MCP gateways to extend their capabilities.

## All Capabilities at a Glance

<CardGroup cols={2}>
  <Card title="Web Search" icon="magnifying-glass" href="./web-search">
    Search the internet for information. AI-optimized results with relevance scoring.

    **Scope:** Agent & MCP
  </Card>

  <Card title="Code Runner" icon="code" href="./code-runner">
    Execute Python and JavaScript code instantly. Perfect for calculations and data processing.

    **Scope:** Agent & MCP
  </Card>

  <Card title="Sandbox" icon="box" href="./sandbox">
    Full compute environment with package installation. Run complex workloads.

    **Scope:** Agent & MCP
  </Card>

  <Card title="HTTP Client" icon="globe" href="./http-client">
    Make HTTP requests to any API. Full auth support (OAuth, Bearer tokens).

    **Scope:** Agent & MCP
  </Card>

  <Card title="Files" icon="file" href="./files">
    Read and write files in your session. Scoped storage for security.

    **Scope:** Agent & MCP
  </Card>

  <Card title="Knowledge Retrieval" icon="book" href="./knowledge-retrieval">
    Vector search over documents. RAG-powered semantic search with embeddings.

    **Scope:** Agent & MCP
  </Card>

  <Card title="Browser" icon="browser" href="./browser">
    Web rendering, screenshots, structured data. Visual web scraping.

    **Scope:** Agent & MCP
  </Card>

  <Card title="Computer" icon="computer" href="./computer">
    Full computer use capabilities. Screenshots, mouse, keyboard control.

    **Scope:** Agent Only ⚠️
  </Card>
</CardGroup>

## Exposure Scopes Explained

### Agent & MCP

Available to both:

* **Agents** - Used in agent-powered interactions and automations
* **MCP Gateways** - Exposed via the Model Context Protocol to MCP clients

Most capabilities are in this category.

### Agent Only ⚠️

Available ONLY to agents:

* **Computer** - Too sensitive for MCP gateway exposure
* Cannot be attached to MCP gateways
* Reserved for direct agent control
* Enforced at multiple layers:
  * Code-level type system
  * Runtime gateway validation
  * UI/form restrictions
  * Per-capability guards

## Capability Features Matrix

| Capability          | Real-time | Stateless | Requires Config | Rate Limited |
| ------------------- | --------- | --------- | --------------- | ------------ |
| Web Search          | ✓         | ✓         | -               | ✓            |
| Code Runner         | ✓         | ✓         | -               | ✓            |
| Sandbox             | ✓         | ✓         | -               | ✓            |
| HTTP Client         | ✓         | ✓         | Optional        | ✓            |
| Files               | ✓         | ✗         | -               | -            |
| Knowledge Retrieval | ✓         | ✓         | Recommended     | ✓            |
| Browser             | ✓         | ✓         | -               | ✓            |
| Computer            | ✓         | ✗         | -               | ✓            |

## Common Capability Combinations

### Minimal Setup

Perfect for simple tasks:

* Web Search
* Code Runner
* Files

### Standard Setup

Recommended for most use cases:

* Web Search
* Code Runner
* Files
* HTTP Client
* Browser

### Full-Featured

Maximum capability:

* All public capabilities
* Plus REST connectors
* Plus plugins

### Knowledge-Heavy

For document analysis:

* Knowledge Retrieval
* Browser (for web docs)
* Files (for local docs)
* Code Runner (for analysis)

## Attaching Capabilities

### To an Agent

1. Open **Agents** > Select Agent
2. Click **Settings** > **Capabilities**
3. Click **Attach Capability**
4. Select **Built-in Capabilities**
5. Choose capabilities
6. Click **Save**

### To an MCP Gateway

1. Open **Gateways** > Select Gateway
2. Click **Capabilities** tab
3. Click **Attach Capability**
4. Select **Built-in Capabilities**
5. Choose capabilities (excludes Agent-only ones)
6. Click **Save**

## Cost Implications

For current pricing details, see [Pricing](https://noorle.com/pricing/).

<Tip>
  Monitor usage in your dashboard to track costs per capability.
</Tip>

## Capability Limits

| Capability          | Key Limit            | Notes                                    |
| ------------------- | -------------------- | ---------------------------------------- |
| Web Search          | 30s timeout          | 10 results per query                     |
| Code Runner         | 128MB memory         | 30s timeout, no network                  |
| Sandbox             | XS–XL sizes          | 20 min idle timeout, 60 min max lifetime |
| HTTP Client         | 2MB request/response | 30s default, 300s max timeout            |
| Files               | 10MB per file        | 100 files per scope                      |
| Knowledge Retrieval | 10GB per index       | 500-char query max                       |
| Browser             | 30s timeout          | 1280×720 default viewport                |
| Computer            | x2–x32 sizes         | Agent-only, 30s SSH timeout              |

## Per-Capability Documentation

Ready to dive deeper? Each capability has detailed documentation:

* **[Web Search](/docs/run/capabilities/web-search)** - Search modes, relevance scoring, news vs general
* **[Code Runner](/docs/run/capabilities/code-runner)** - Python & JavaScript execution, input/output
* **[Sandbox](/docs/run/capabilities/sandbox)** - Packages, shell, environment setup
* **[HTTP Client](/docs/run/capabilities/http-client)** - Authentication, headers, status handling
* **[Files](/docs/run/capabilities/files)** - Read/write operations, scopes, path handling
* **[Knowledge Retrieval](/docs/run/capabilities/knowledge-retrieval)** - Embeddings, semantic search, RAG
* **[Browser](/docs/run/capabilities/browser)** - Web rendering, screenshots, structured data
* **[Computer](/docs/run/capabilities/computer)** - Screen control, mouse, keyboard automation

## Quick Recipes

### Search and Summarize

Use Web Search + Code Runner:

1. Web Search retrieves recent articles
2. Code Runner processes results
3. Agent summarizes findings

### Calculate and Store Results

Use Code Runner + Files:

1. Code Runner does computation
2. Files stores results
3. Agent can retrieve later

### Document Analysis

Use Knowledge Retrieval + Browser:

1. Browser fetches documents from web
2. Knowledge Retrieval searches content
3. Files stores analysis results

## Next Steps

* [Enable Web Search](/docs/run/capabilities/web-search)
* [Run Code](/docs/run/capabilities/code-runner)
* [Create Agents](/docs/run/agents/creating-an-agent)
* [Build REST Connectors](/docs/run/connectors/rest-connectors)