MSCI MCP Server
Conversational AI access to MSCI's proprietary index data, factor exposures, constituents, and methodologies for licensed MSCI clients.
The MSCI connector is a provider-hosted MCP integration from MSCI Inc. that gives investment professionals conversational access to MSCI's proprietary data and analytics. The first capability shipped on the platform is MSCI IndexAI Insights, a natural-language interface to more than 68,000 standard and custom MSCI indexes with up to 20 years of performance history. Users can ask plain-language questions about index performance, country and sector exposures, factor exposures, constituent holdings, and index methodologies, and get answers grounded directly in MSCI's official data and documentation.
The connector was announced in February 2026 and is available inside Anthropic's Claude (web, desktop, mobile, Claude Code, and the Claude API) as well as through the MSCI app for OpenAI ChatGPT. MSCI has described the integration as the foundation for a wider series of AI-enabled offerings, with planned extensions into Private Assets workflows such as due diligence, underwriting, analytics, and reporting. Although MSCI is widely associated with ESG ratings and Barra factor risk models, the current launch is centered on index data and methodology Q&A; broader ESG ratings and risk analytics coverage has been signaled as part of the roadmap rather than confirmed in the initial release.
Access is gated by existing MSCI entitlements. Each user authenticates individually, and Claude inherits the user's MSCI permissions, so the connector can only surface data the user is already licensed to see. There is no public GitHub repository and no self-hosted version; the server is operated by MSCI and consumed as a managed connector.
Tools
| Tool | Description |
|---|---|
index_performance_query |
Answer natural-language questions about MSCI index performance, including returns over arbitrary periods and side-by-side comparisons across indexes, with charts and tables. |
index_constituents_query |
Retrieve constituent holdings and weightings for an MSCI index, including top constituents and breakdowns refined through follow-up questions. |
index_exposures_query |
Surface country, sector, and factor exposures for MSCI indexes to support portfolio construction and attribution analysis. |
index_methodology_query |
Return methodology details and construction rules for MSCI indexes, grounded in MSCI's official methodology documents. |
executive_takeaways |
Generate concise, decision-ready summaries highlighting key differences between indexes or across time periods. |
Prerequisites
- An active MSCI license that covers the index data you want to query (the connector is available exclusively to clients licensed for MSCI index solutions).
- A Claude account (free or paid) or ChatGPT account, depending on which client you want to use.
- Individual MSCI credentials. Each user authenticates separately; Claude inherits the user's MSCI permissions.
Setup in Claude
- Open Claude in the desktop app or at
https://claude.ai/and sign in. - From the top menu, click File, then Settings to open the configuration panel.
- Under Capabilities, select Connectors.
- Locate the MSCI connector in the directory and click the
+button to add it. - Complete the OAuth login flow with your MSCI credentials to authorize the connection.
The MSCI connector works across Claude web, Claude desktop, Claude mobile, Claude Code, and the Claude API.
Setup in ChatGPT
In ChatGPT, install the MSCI app from the apps/connectors directory and sign in with your MSCI credentials when prompted.
Notes
- There is no public MCP server URL or self-hosted distribution; MSCI operates the server.
- There is no GitHub repository and no npm or PyPI package to install.
- Admins on Claude Enterprise / Claude Cowork can authorize the MSCI connector centrally and bundle it into team plugins.
- Ask plain-language questions about MSCI index performance and get charts and tables comparing returns across periods.
- Pull top constituents and weightings for a specific MSCI index during portfolio review and refine results with follow-up questions.
- Investigate country, sector, and factor exposures of an MSCI index to support attribution and risk discussions.
- Look up MSCI index methodology rules without leaving the chat, grounded in MSCI's official documentation.
- Generate executive-ready summaries highlighting key differences between two or more MSCI indexes.
- "Compare the trailing 1-year and 5-year returns of MSCI ACWI and MSCI World, and show me a chart."
- "What are the top 10 constituents of the MSCI Emerging Markets Index and their weights?"
- "Break down the sector and country exposures of the MSCI EAFE Index."
- "Summarize the methodology behind the MSCI USA Quality Index."
- "Give me an executive summary of the key differences between MSCI World and MSCI World ex USA over the last 3 years."
- Built and operated by MSCI directly, so answers are grounded in official MSCI index data and methodology documents.
- Broad coverage of more than 68,000 standard and custom MSCI indexes with up to 20 years of history.
- Available across Claude (web, desktop, mobile, Code, API) and ChatGPT, with admin controls for enterprise rollout.
- Inherits the user's MSCI entitlements, so existing data licensing and permissions are respected.
- Restricted to clients with an active MSCI index license; not usable by non-subscribers.
- Initial release is focused on index data (IndexAI Insights); ESG ratings, Barra factor models, and broader risk analytics are signaled on the roadmap but not all confirmed as live tools yet.
- No public technical documentation, MCP server URL, or self-hosted option; clients are dependent on MSCI's managed connector.
- FactSet MCP connector for portfolio and market data inside Claude.
- Snowflake Cortex MCP server, which can surface third-party data sets (including MSCI) for licensed Snowflake customers.
- Financial Datasets MCP server (
financial-datasets/mcp-server) for open access to public equity fundamentals and market data, as a lighter, non-MSCI alternative.