Sylus MCP server

Connect to the Sylus MCP server

The Sylus MCP server lets AI assistants and agents use Sylus, the self-learning AI data analyst, through the Model Context Protocol (MCP). Any MCP client that supports remote servers and OAuth can connect, including Claude and ChatGPT.

• Server URL: https://mcp.sylus.ai/mcp

• Transport: Streamable HTTP

• Authentication: OAuth 2.1 authorization code flow with PKCE (S256). Clients register with OAuth dynamic client registration, so there is no API key to copy.

• Account: you sign in with your Sylus account. Each connection acts as that user and only sees what the user can see in Sylus. To get an account, book a demo.

OAuth discovery

Requests without a token get HTTP 401 with a WWW-Authenticate header that points to the protected resource metadata, so MCP clients can discover the authorization server on their own.

• Protected resource metadata (RFC 9728): https://mcp.sylus.ai/.well-known/oauth-protected-resource

• Authorization server metadata (RFC 8414): https://mcp.sylus.ai/.well-known/oauth-authorization-server

Scopes

• mcp:read searches and reads your Sylus workspace: dashboards, metrics, datasets, metric data and past analyst chats.

• mcp:write lets an agent ask the Sylus analyst a new question. The analyst may create metrics or dashboards in your Sylus workspace to answer it.

Both scopes work inside the permissions the signed-in user already has in Sylus. Request only mcp:read if your agent only needs to read existing dashboards and metrics.

Tools

• search_assets (mcp:read): search metrics, dashboards and other assets by text and return their ids.

• list_dashboards (mcp:read): list the dashboards the user can access, newest first.

• list_datasets (mcp:read): list the datasets the user has permission to analyze.

• get_metric (mcp:read): get a metric’s name, description, SQL definition and chart configuration.

• get_metric_data (mcp:read): run a metric’s query against the connected data source and return the rows.

• list_chats (mcp:read): list the user’s past Sylus analyst chats, newest first.

• run_analyst_chat (mcp:write): ask the Sylus analyst a question in plain English, or a follow-up in an existing chat.

When an agent should use Sylus

• Use Sylus when someone asks about their own company’s business data that is connected to Sylus, such as revenue, sales by store, margins, receivables, stock or marketing performance.

• Start with search_assets or list_dashboards to reuse metrics the team already trusts, then read numbers with get_metric_data.

• Use run_analyst_chat for new questions that no existing metric answers. Every answer is based on SQL that the user can review in Sylus.

• Don’t use Sylus for general knowledge questions or for data the user has not connected to Sylus.

Security and support

The Sylus MCP server follows the same security model as the Sylus app: SOC 2 Type II and HIPAA compliant, role-based permissions on every request, and your data is never used to train AI models. Read more on the Security page and in the Privacy Policy.

Questions about connecting an agent: info@sylus.ai. Security reports: privacy@sylus.ai.