Model Context Protocol (MCP)¶

The Model Context Protocol (MCP) is an open standard that enables AI assistants to securely connect to external data sources, tools, and services. It defines a common language for how AI models discover and call capabilities exposed by external systems without hardcoding integrations or writing custom connectors for each tool.

How MCP Works¶

MCP follows a client-server architecture:

• MCP Server: An application (like Qualytics) exposes a set of tools and resources through the MCP protocol. The server handles authentication, executes the tool logic, and returns structured results.
• MCP Client: An AI assistant or agent (like Claude Desktop, ChatGPT, or Cursor) connects to one or more MCP servers and can call their tools during a conversation.
• LLM: The language model decides which tools to call, in what order, and how to interpret the results. The MCP protocol itself is model-agnostic.

When you send a message to an MCP-connected assistant:

1. The LLM receives your message along with a description of available tools.
2. It decides whether to call a tool and which one.
3. The client sends the tool call to the MCP server.
4. The server executes the action and returns a structured result.
5. The LLM receives the result and incorporates it into its response.
6. This loop repeats until the LLM produces a final answer.

Through this design, the LLM does not connect directly to your data infrastructure. It receives the context and structured tool results needed for the request. The data a tool can return is governed by that tool's capabilities and the requesting user's Qualytics permissions. Most results describe your data rather than quote it. A few tools can surface individual values, such as a value quoted in a quality check failure explanation or a profile statistic, but no tool returns rows or records in bulk, and masked fields stay masked.

Key Concepts¶

Tools¶

Tools are the primary mechanism through which MCP servers expose functionality. Each tool has:

• A name: how the LLM references it (e.g., query_qualytics)
• A description: natural language description the LLM uses to decide when to call it
• An input schema: the parameters the tool accepts (JSON Schema)
• An output: structured data returned after execution

The LLM reads tool descriptions and schemas to understand what each tool does and how to call it correctly. Well-written tool descriptions are essential because they allow the model to select the right tool for a given user request.

Resources¶

MCP also defines a resources concept for exposing read-only data (files, database records, documents) that the LLM can reference as context. Qualytics primarily uses tools rather than resources for its MCP integration.

Prompts¶

MCP supports server-defined prompt templates: pre-built message structures that guide the LLM through specific workflows. Support for prompts depends on the client and server.

Tool Registration¶

MCP supports dynamic tool discovery, allowing clients to start with a small core set and load additional tools on demand. Qualytics' MCP server takes a different approach. It registers its full tool catalog upfront, and tool descriptions are kept concise so the complete catalog fits within the LLM's context budget.

Transport¶

MCP supports multiple transport mechanisms. Qualytics exposes its MCP server over streamable HTTP, the standard transport for remote MCP servers, which uses Server-Sent Events (SSE) to stream responses. This means:

• No special local software is required to connect external clients.
• External clients authorize through OAuth. The client sends a separate, short-lived MCP access token with each request.
• Streaming responses are supported for long-running tool executions.

What You Can Do with MCP + Qualytics¶

Quality Trends & Insights¶

Analyze data quality trends over time to understand how your quality posture is evolving:

• "How has the quality score for our orders table changed over the past month?"
• "Show me the trend of anomaly volume across all datastores"
• "What are the top anomaly drivers in our production database?"

The AI retrieves time-series quality metrics, operation execution insights, and trend analysis to help you identify patterns and measure improvement.

Global Search & Asset Discovery¶

Search across your entire data landscape to find specific assets, tables, fields, or quality checks:

• "Find all tables related to customer data across all our datastores"
• "Search for any containers named 'transactions'"
• "What quality checks mention revenue?"

Operational Actions¶

Trigger and monitor data operations, manage tags, send notifications, and create tickets through conversation:

• "Run a profile operation on the customers table"
• "Tag the orders and transactions tables as 'finance-critical'"
• "Send a notification to the data-engineering Slack channel about the quality issue"
• "Create a Jira ticket for the null values in the merchant_id field"

For the complete list of available tools, and the category of data each one shares with the model, see the Tool Catalog.

Why MCP Matters for Data Quality¶

Traditional integrations between AI tools and data platforms require custom-built connectors, hardcoded API calls, and constant maintenance as APIs evolve. MCP standardizes this so any MCP-compatible AI client can connect to any MCP server and immediately discover its capabilities.

For data quality workflows, this means:

• Portability: The same Qualytics MCP server works with Claude Desktop, ChatGPT, Cursor, and any future MCP-compatible tool, without changes.
• Security: The MCP server enforces authentication and RBAC on every tool call. The LLM receives results through authenticated, permission-aware tools rather than connecting directly to your databases.
• Auditability: Every tool call is a discrete, logged action. You can see exactly what the AI did and why.
• Composability: Complex workflows emerge naturally from combining simple tools, driven by the LLM's reasoning rather than hardcoded orchestration logic.

Frequently Asked Questions¶

Can I connect external AI clients to Qualytics?¶

Yes. Connect a compatible client to your Qualytics MCP URL and complete the browser authorization with a Personal API Token. The client receives a separate MCP token. AgentQ is a built-in assistant and does not need this connection. See Connecting External AI Clients for setup steps.

Which external AI clients are supported?¶

The guide includes setup steps for these clients:

Other clients that support remote MCP over Streamable HTTP and browser OAuth can also connect. See Connecting External AI Clients.

Do external clients have the same capabilities as AgentQ?¶

They discover the same Qualytics tool catalog, subject to the client's granted scopes and the user's Qualytics permissions. AgentQ and external clients are separate applications. Each client has its own model, conversation history, interface, and handling of prompts and tool results. AgentQ's page context, suggestions, and PDF export belong to its built-in chat.

What is the difference between AgentQ, the MCP server, and agentic endpoints?¶

AgentQ and MCP clients use the same tool catalog. They manage conversations and authorization separately.

What tools are available?¶

The available tools depend on your Qualytics permissions and the relevant feature configuration. See the Tool Catalog for the authoritative list and the category of data each tool shares.

Can the MCP server write to my source datastores?¶

No. Tools can change Qualytics assets and can start operations that write to an enrichment datastore. They do not write back to source datastores. Review the actions your client proposes before approving them.

How is MCP access authorized?¶

Your MCP client opens a Qualytics authorization page where you enter a Personal API Token. Qualytics grants the client a separate MCP access token and checks the source token's identity and permissions on tool calls. The client can request read-only access; when it requests no scopes, the authorization page shows read and write access. Revoking the Personal API Token ends the authorization. See Connecting External AI Clients for reconnection and expiry details.

See Also¶