RyxelData MCP Server: Connect AI to Live Financial Data 2026
The RyxelData MCP server enables connections between AI assistants such as Claude, Cursor, and Claude Code and real-time institutional financial datasets within a chat. Using the Model Context Protocol, one can use an AI can query insider transactions, 13F institutional holdings, proposed sales, fund compositions, and supply chain data in real-time without any middleware involved. This article will help you understand how the MCP works and how to connect to it using different AI assistants, along with examples of what information the different data sets provide to develop an AI trading strategy.
What is the RyxelData MCP server and how does it work?
Definition The RyxelData MCP server is a Model Context Protocol endpoint at mcp.ryxel.ai that gives AI assistants real-time, authenticated access to institutional financial datasets via structured tool calls. It supports Claude, Cursor, Claude Code, and OpenCode.
Model Context Protocol is an open standard that allows AI assistants to call external data sources as tools, mid-conversation and mid-reasoning, without leaving the interface. Rather than requiring a developer to pre-fetch data and paste it into a prompt, the AI determines what to query, calls the relevant MCP tool, receives clean structured output, and reasons over it immediately. The model retains control of the query, and the data arrives normalized and ready for analysis.
RyxelData's MCP server exposes every dataset in an active subscription as a callable tool. That includes insider transactions parsed from Form 4 filings, institutional holdings from 13F submissions, proposed sales filed under SEC Rule 10b5-1, mutual fund and ETF composition data, and supply chain relationship maps between public companies. All data is returned as consistent, normalized JSON drawn from the same primary institutional sources used by hedge funds, formatted specifically for language model consumption.
Schema consistency is a critical consideration in this context. An MCP tool call that returns inconsistent field names across historical records, or ambiguous null values, produces downstream reasoning errors that are difficult to trace and compound at machine speed. RyxelData's datasets maintain consistent field naming across more than 20 years of records, encode null values explicitly, pre-apply corporate actions, and paginate responses within context window limits. That architecture removes the failure modes that make most financial data unreliable in agentic AI workflows.
How to connect the RyxelData MCP server
Connecting the RyxelData MCP server takes under 10 minutes on any supported assistant. Two prerequisites apply across all four platforms: a Ryxel API key generated from Dashboard > API Keys, and at least one active dataset subscription from Dashboard > Subscriptions. The MCP server authenticates user identity but exposes tools only for datasets with an active subscription. If a tool does not appear after connecting, subscription status should be verified first.
Connect via Claude.ai
1 | Open Claude.ai and navigate to Settings > Connectors. |
2 | Click Add a custom connector and enter the MCP URL: mcp.ryxel.ai |
3 | Save. RyxelData will appear in your connectors list. |
4 | Click Connect to trigger the OAuth flow and authenticate with your API key. |
5 | Once authenticated, Ryxel tools are live in every Claude conversation. |
Connect via Cursor
1 | Open Cursor and go to Settings → Tools & MCP |
2 | Click New MCP, this opens your |
3 | Add the Ryxel server to the |
4 | Save the file and go back to Settings → Tools & MCP |
5 | Click Connect next to the Ryxel server to trigger the OAuth flow and authenticate with your API key |
Connect via Claude Code
1 | Run the following command in your terminal:
|
2 | Open a Claude Code session and type |
3 | Verify the server is connected at any time:
To add it for a specific project only, pass |
Connect via OpenCode
1 | Add the MCP server:
|
2 | Authenticate:
This will open a browser window and redirect you to the Ryxel authentication page where you can enter your API key. Once authenticated, the Ryxel tools will be available in OpenCode. |
The five live datasets: what each one gives your AI
Each MCP tool maps to one of RyxelData's five live datasets. Tool access depends entirely on which datasets are active in a given subscription.
Insider Transactions: Form 4 filings, real-time
Insider transactions represent the most direct behavioral signal available in public equity markets. When a CFO, board member, or major shareholder purchases shares on the open market, spending personal capital rather than exercising options, that represents internal conviction that no press release can replicate. RyxelData parses Form 4 filings from SEC EDGAR in real time and delivers them as normalized JSON. An AI querying this tool can cluster executive buy patterns across a watchlist, flag accumulation against declining prices, and surface companies where insider conviction is accelerating well before it appears in analyst coverage.
Institutional Holdings: Aggregated 13F positions
Asset managers with more than 100 million dollars in assets under management are required to file 13F disclosures of their equity positions on a quarterly basis. RyxelData normalizes those submissions into structured holdings data that shows exactly where large institutional capital is positioned. For AI strategy systems, this dataset serves two purposes: identifying crowding risk in sectors with heavy hedge fund concentration that are vulnerable to macro shocks, and tracking quarter-over-quarter position changes that frequently precede price movements by several weeks.
Proposed Sales: Insider intent before execution
Under SEC Rule 10b5-1, insiders must file planned sales before they execute, creating a public record of intent before any price impact occurs. Proposed sales and completed insider transactions answer different questions. Proposed sales indicate where insider conviction is pointing before the market sees any transaction. The timing window between the filing of intent and actual execution is the primary analytical value. Most data providers surface completed transactions only. RyxelData surfaces both.
Funds: ETF and mutual fund profiles with holdings
Mutual fund and ETF composition data shows what percentage of a company's float is held in passive vehicles. This dimension changes how volume spikes and price action should be interpreted. A company where 60 percent of the float sits in passive ETFs responds to index rebalancing events very differently from a company with dispersed active ownership. For AI agents building a complete picture of positioning at a given name, this dataset fills in the passive-vehicle layer.
Supply Chain: Company relationship maps
The supply chain dataset maps documented customer, supplier, and partner relationships between public companies. It is the dataset that enables second-order analysis. If a major supplier shows unusual insider selling alongside declining institutional position size, every downstream customer in that relationship graph inherits a portion of that risk signal. An AI with access to supply chain data can model contagion pathways that do not appear in any individual company's filing. That type of analysis previously required a research team and hours of manual cross-referencing. With RyxelData, it is available as a single tool call.
Key Fact: Querying all five datasets at a single company within the same session produces convergence signals that are structurally more reliable than any individual dataset alone. RyxelData's normalized schema makes cross-dataset joins possible without any transformation layer.
Building a Reliable AI Market Strategy with RyxelData
Having a live API connection is the starting point, not the strategy. What determines whether your workflow produces reliable output is the architecture built on top of it.
A production-grade AI market strategy using RyxelData requires five components working together:
- Defined signal hierarchy: Insider transactions as primary conviction signals, 13F holdings as crowding filters, proposed sales as a timing layer, supply chain data as a risk propagation monitor, and funds data as a float-composition reference.
- Query cadence matched to each dataset: Insider data warrants daily queries, while 13F holdings update quarterly and querying them more frequently adds no value.
- Convergence logic: The strongest outputs come when multiple datasets reinforce the same signal simultaneously, so what counts as divergence, and partial convergence must be explicitly defined.
- Regime context as a modifier: Insider buying during a sector correction carries different weight than the same signal in a bull market, and that distinction must be built into the architecture.
- Historical validation across at least two full market cycles: Before any live deployment, using RyxelData's 20+ years of historical depth.
RyxelData dataset pricing
All datasets are available individually with no bundling requirement. The all-in-one bundle at $149 per month covers every live dataset plus all seven upcoming datasets as they roll out, including Financial Statements, Earnings, Dividends, IPO Calendar, and Splits.
Dataset | Insider transactions | Institutional holdings | Proposed sales | Funds | Supply chain | All-in-one Bundle |
|---|---|---|---|---|---|---|
Ryxel | $49/mo | $19/mo | $29/mo | $59/mo | $129/mo | $149/mo |
All 5 live datasets plus 7 upcoming soon are available through RyxelData's API with per-dataset pricing, so you only pay for what you need.
Getting started
The RyxelData MCP server is available. Setup takes under 10 minutes on Claude, Cursor, Claude Code, or OpenCode. Creating an account at ryxel.io, subscribing to the relevant datasets, generating an API key from Dashboard > API Keys, and adding the MCP URL as a custom connector are the only steps required. Once the OAuth flow completes, live institutional data is accessible in every session. Full documentation is available at ryxel.io/docs. MCP-specific setup and tool references are at ryxel.io/docs/mcp-server.
Available Tools
Once connected, your AI assistant can access the following tools:
Tool | Description |
|---|---|
get_insider_transactions | List Insider Transactions |
get_institutions_holdings | Detailed Institutional Holdings |
get_institutions_issuers | Issuer-Aggregated Holdings |
get_institutions_managers | Manager-Aggregated Holdings |
get_proposed_sales | List Proposed Sales (Form 144) |
get_funds_holdings | Fund Holdings |
get_funds_profile | Fund Profile |
get_funds_list | List Funds |
get_supply_chain | Supply Chain Relationships |
get_companies | List Companies |
Access to each tool depends on your active subscriptions.
Frequently asked questions
What is the RyxelData MCP server?
The RyxelData MCP server is a Model Context Protocol endpoint at mcp.ryxel.ai that connects AI assistants to live institutional financial datasets through authenticated tool calls. It supports Claude, Cursor, Claude Code, and OpenCode. Once connected, an AI can query insider transactions, 13F holdings, proposed sales, funds, and supply chain data directly inside the conversation, with no custom API integration required.
How do I connect RyxelData to Claude?
To connect RyxelData to Claude, navigate to Claude.ai Settings > Connectors > Add a custom connector, enter mcp.ryxel.ai save, and click Connect to complete OAuth authentication with the Ryxel API key. Ryxel tools become available in every Claude conversation immediately after authentication. An active dataset subscription is required for tools to appear.
Is RyxelData suitable for AI agents and LLM-powered financial applications?
RyxelData was built specifically for AI-native workflows. Its datasets maintain consistent field naming across more than 20 years of records, encode null values explicitly, pre-apply corporate actions, and return paginated JSON scoped for context window consumption. These design decisions allow Ryxel data to be consumed directly by language models without a normalization or transformation layer, which is the primary reason it performs reliably in agentic AI workflows where general-purpose financial data providers often introduce errors.
What is the difference between RyxelData's proposed sales and insider transactions datasets?
Insider transactions record completed buys and sells after they execute and are reported through Form 4 filings. Proposed sales record insider intent to sell before execution, filed under SEC Rule 10b5-1. The key distinction is timing: proposed sales give AI agents a forward-looking signal window between the filing of intent and the actual market transaction, which completed transaction data cannot provide. Both are available as separate MCP tools and produce the most value when queried together.
How does RyxelData compare to FactSet and LSEG for developers?
FactSet and LSEG are enterprise platforms built for large institutional teams with dedicated data engineering resources and procurement budgets. Both require account representative access, proprietary interfaces, and significant contractual overhead. RyxelData draws from the same primary data sources, including SEC EDGAR, regulatory filings, and exchange feeds, but delivers them through a clean REST and JSON API with a single Bearer token, no SDK requirement, and per-dataset pricing starting at $19 per month. For independent developers and fintech teams, RyxelData is the practical alternative.
Does RyxelData have an MCP server?
Yes. RyxelData operates a native MCP server at mcp.ryxel.ai, compatible with Claude, Cursor, Claude Code, and OpenCode. Setup requires a Ryxel API key and at least one active dataset subscription.