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

# Exposed MCP Tools

> A comprehensive breakdown of all capabilities exposed to connected MCP clients.

When an MCP client initializes a handshake with Arivu, it registers several modular capabilities directly accessible to the host LLM context layer.

<CardGroup cols={2}>
  <Card title="arivu_query" icon="magnifying-glass">
    Executes natural language queries dynamically converting intent into safe SQL.
  </Card>

  <Card title="arivu_get_schema" icon="diagram-project">
    Returns the structural table configurations and column contexts instantly.
  </Card>
</CardGroup>

## The Tool Registry

Here is the exact suite of tools transmitted during the initialization phase:

### 1. Data Interactions

* `arivu_query(question: str)`: The core pipeline router. Parses a textual question, interfaces with LangGraph, and drops back a heavily verified answer.
* `arivu_execute_sql(sql: str, max_rows: int)`: Executes raw SQL directly against the database, bypassing the Arivu safety pipeline. Destructive operations run immediately.
* `arivu_get_session_history(session_id: str)`: Pulls prior conversation context for the designated thread from Arivu memory safely.

### 2. Admin & Approvals

If your configuration includes `ARIVU_DB_MODE="user"`, destructive executions triggered via `arivu_query` will hit an approval queue. MCP clients can view and action these queues natively.

* `arivu_get_pending_approvals()`: Lists all operations globally waiting for admin authorization.
* `arivu_approve(session_id: str)`: Approves the pending SQL mapped to the session, forcing execution successfully.
* `arivu_reject(session_id: str)`: Hard rejects the destructive operation, reverting pipeline state.

### 3. Diagnostics & Cache

* `arivu_get_schema()`: Retrieves the current database schema including tables, columns, types, and foreign keys.
* `arivu_refresh_schema()`: Triggers a hard re-extraction of backend SQL schema bypassing the set cache TTL. Use this when the underlying database is actively migrating.
* `arivu_test_connection()`: Provides an instantaneous health check verifying dialect networking availability, schema load status, and latency.
* `arivu_list_dialects()`: Informs the host LLM regarding what dialects the current environment successfully installed (`pymysql`, `psycopg2`, etc).