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

# Agent Studio

**Overview:** Quantum Heaps features an integrated Agent Studio that enables the creation of specialized sub‑agents. Each sub‑agent is purpose‑built for a specific task, with its own instructions, capabilities, and data access. This ensures that the agent operates with dedicated focus on the configured objective, rather than functioning as a general‑purpose assistant. 

<br />**What Agent Studio Is Used For**

Use Agent Studio to:

* Create a special agent focused on one job — like ICP discovery, lead scoring, or meeting prep
* Start from a pre-built template instead of writing instructions from scratch
* Control exactly which CRM modules and permissions each agent can access
* Manage a library of agents — published, draft, or deactivated

#### **Your Agents**

Below the templates, **Your Agents** shows every agent you've created as a card — name, description, number of configured sections, scope (e.g., "web on," "crm-scoped"), and when it was last edited. Active agents show a **Deactivate** button directly on the card. Click **+ Create special agent** to start a new one from scratch or from a template.

<Frame>
  <img src="https://mintcdn.com/quatumheaps/-jQYdYtpsW7ToMeo/images/image-1.png?fit=max&auto=format&n=-jQYdYtpsW7ToMeo&q=85&s=3f18544d6fffa202eabced418b0188e1" alt="Image" width="1460" height="887" data-path="images/image-1.png" />
</Frame>

<br />

### **Building a Special Agent**

Click **+ Create Special Agent**, or open an existing agent, to configure it across several sections:

#### **Instructions**

Write what the agent should do when it runs — Quantum Heaps supports a detailed system prompt (up to 8,000 characters) where you typically define:

* **Core Responsibilities** — the numbered steps the agent follows (e.g., Define, Validate, Discover, Dedupe, Refresh)
* **Data Access** — what it can read vs. write, and where
* **Response Format** — how it should structure its output

#### **Knowledge Base**

Upload reference files — PDFs, docs, sheets, or images — that the agent should always have available, on every run. Click **Add reference files** or drag and drop directly onto the upload zone.

<br />

<br />

<br />

<br />

<br />