Oxford Economics MCP Server

Overview

The Oxford Economics MCP Server is a remote Model Context Protocol (MCP) endpoint that lets AI assistants query Oxford Economics data directly. Clients such as Claude (Anthropic) and ChatGPT (OpenAI) can connect to the server as a tool provider and retrieve economic data and research on a subscriber's behalf, scoped to that subscriber's entitlements.

Access is restricted to entitled Oxford Economics subscribers and approved partners.

What you can do with it

The server exposes two purpose-built tools.

EconomicData

Answers economic questions that require numerical data, drawn from the Oxford Economics databank.

Best for: single-indicator queries (e.g. "What is the GDP of the UK?") and cross-region comparisons of the same indicator (e.g. "GDP growth for the US, UK, Germany, and France")
Returns: natural-language answers, typically including data values and markdown tables
Limit: up to 10 datapoints per query; use multiple calls for larger needs
Tip: keep questions focused on a single indicator — avoid bundling unrelated metrics into one query

EconomicAnalysis

Answers economic questions that require interpretation, narrative, or conclusions, drawn from Oxford Economics reports and research.

Best for: explanatory questions where prose summaries or report-derived analysis are expected rather than raw numbers
Returns: textual explanations based on research published in the last 4 months by default
Tip: you can request a specific date range in your question to filter results

Inputs (both tools)

Input Type Description question string The economic question, kept simple and focused language string ISO 639 language code for the response (e.g. en) conversationId string (GUID) Optional identifier to group related queries in a session

Getting access

To use the MCP server you need:

An Oxford Economics subscription that includes the MCP entitlement
An API key (or OAuth credentials for clients that support it)

Contact your Oxford Economics account manager to confirm your entitlement and obtain credentials.

Connecting your AI client

Connection details

Item Value Server URL https://services.oxfordeconomics.com/mcp Protocol MCP over HTTPS (JSON-RPC 2.0), MCP protocol version 2025-03-26 Transport security TLS 1.2 / TLS 1.3 Authentication API key header, or Oxford Economics sign-in through supported agent applications (OAuth 2.0)

Step-by-step

Confirm your Oxford Economics agreement includes the MCP entitlement
Obtain an API key from your account manager, or confirm your client is supported for sign-in (see below)
Add the server to your AI client using the instructions for your platform (see below)
Authenticate with your API key or by completing the OAuth sign-in flow
Verify the connection by asking a simple question, e.g. "What is the latest GDP figure for the UK?"

Supported platform setup guides

Claude (Anthropic):
- Find "Oxford Economics" in the Anthropic & Partners Connection Directory.
- For help go to https://support.claude.com/ Search for: "Use connectors to extend Claude's capabilities".
ChatGPT (OpenAI):
- For help go to https://help.openai.com/ Search for: "Developer mode and MCP apps in ChatGPT", and follow their instructions to add a custom app.
Other MCP-compatible clients:
- Configure with the Server URL above and choose an available auth method.
- For OAuth, contact your account manager to discuss enabling Dynamic Client Registration for your chosen application.

What data you can access

The server enforces your subscription entitlements on every request — you will only ever receive data you are licensed for.

EconomicData draws on the Oxford Economics Global Macro and Global Industry databanks. Access is granted at the databank level: if your agreement includes a databank, you can query any region or variable within it.

EconomicAnalysis draws on the Oxford Economics research library (MyOxford). Access is enforced at the individual report level, matched against your research entitlements.

Authentication

API key: supply your Oxford Economics API key in the Api-Key request header. Keys are validated against your account and entitlements on every request.

OAuth 2.0: supported for interactive clients, using Authorization Code flow with PKCE. Refresh tokens are available for long-lived sessions. Dynamic Client Registration is supported for approved applications.

Rate limiting

Requests are rate limited per user with both a short-window burst cap and a daily limit. Daily limits are tiered by entitlement level. Exceeding a limit returns a standard "too many requests" response.

Data handling

Answering a query may involve a third-party AI model to interpret the question and compose the response. Oxford Economics applies the following principles:

Grounded in licensed data. Answers are generated from Oxford Economics' own data and research, scoped to your entitlements — not from open-ended model knowledge.
No training on your inputs. The AI model is used under an enterprise agreement under which subscriber inputs and outputs are not used to train the provider's models.
No personal identifiers shared externally. Only the content needed to answer the question is sent to the AI model. Your name, email address, and company identifier are not shared with the AI provider.

Oxford Economics records query text, responses, timestamps, and your internal account identifier for operational support and analytics, in line with our contractual obligations.

For full details on sub-processors, data retention, and security measures, please refer to our Technical and Organisational Measures (TOM), available from your account manager.

Support

For access, entitlement, or connection queries, contact your Oxford Economics account manager or reach out via our standard support channels.