> ## Documentation Index
> Fetch the complete documentation index at: https://acem-52171079.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Prompts

> System prompts to guide agent behavior

The MCP Server includes pre-configured prompts (`app/prompts/`) designed to help Agents understand how to effectively use the DataDot toolset.

## Discovery & Context

### `workspace_overview`

**Purpose**: Contextual awareness.
Provides the agent with a high-level summary of the current workspace, including statistics, key file locations, and recent activity levels. It helps the agent "orient" itself before performing tasks or answering questions.

### `file_inspection`

**Purpose**: File analysis guidance.
Guides the agent through inspecting workspace files. It outlines a workflow for listing all files, finding specific ones by name, and reading content to provide insights or recommendations on file organization and "pinned" status.

## RAG & Retrieval

### `query_sse`

**Purpose**: Retrieval guidance.
Instructs the agent on how to use the streaming RAG capabilities of `query_content`. It emphasizes waiting for the finalized answer, interpreting source citations, and synthesizing information from multiple chunks.

## File Operations

### `upload_process`

**Purpose**: Upload workflow.
A step-by-step guide for uploading and processing documents. It instructs the agent to handle file reception, upload via `upload_file`, optional custom processing, and verification of the upload success.

### `deletion_safety`

**Purpose**: Safety guardrails.
Used as a pre-check before a `delete_file` operation. It acts as a critical safety mechanism, instructing the agent to:

1. Verify the file exists.
2. Explain the irreversible nature of deletion.
3. **Require explicit user confirmation** (often asking them to type the filename) before proceeding.

## Maintenance & Diagnostics

### `embedding_refresh`

**Purpose**: Maintenance workflow.
Guides the process of refreshing workspace embeddings. Because this can be a long-running operation, it instructs the agent to check current status, warn the user about the time required, and verify completion stats.

### `troubleshooting`

**Purpose**: Diagnostic support.
Provides a knowledge base of common issues (e.g., search failures, upload errors) and their solutions. It lists diagnostic tools like `check_embeddings` and `search_files` to help the agent diagnose workspace health problems.

## Using Prompts

Agents typically request these prompts during their initialization phase or when switching contexts to set their system instructions (System Prompt).