Signal8 MCP

What signal8ai/signal8-mcp MCP server does

The signal8ai/signal8-mcp MCP server connects MCP-compatible AI clients to Signal8 data covering public companies, markets, SEC filings, ownership, compliance, and related research. The repository describes 86 live tools, while noting that some documented categories may be disabled and not registered at runtime.

Company tools can search by name or ticker, return enriched profiles, provide current quotes, calculate market metrics, retrieve short interest and float data, and return historical prices or percentage changes. Additional company research includes financial statements, earnings, executives, news, analyst consensus and estimates, clinical trials, and peer-related information where available.

SEC workflows include filing discovery, company- and filing-level filters, raw filing document retrieval, and exhibit listing. Intelligence features cover insider transactions, institutional holdings, counterparties, legal counsel, cross-company patterns, compliance alerts, listing classification, dilution analysis, instruments, and company screening. Market-wide tools cover movers, breadth, news, calendars, trading halts, material filings, and post-earnings performance.

How it works

The package runs as a local MCP process. An MCP client starts the npm command and communicates with the process using the client’s configured MCP connection. The server uses the SIGNAL8_API_KEY environment variable to authenticate requests to Signal8.

Many workflows begin with search_companies, which returns identifiers such as ticker, CIK, name, and exchange. A client can then pass the ticker to company-specific tools. Filing research commonly starts with search_sec_filings; the returned filing ID can be used with get_filing_document or get_filing_exhibits.

Responses come from live Signal8 data and may have different credit costs by tool. The reference material lists examples ranging from one credit for basic company lookup to higher costs for screening, compliance, cross-company intelligence, and aggregated bundles. Exact account limits and pricing are not specified in the supplied material.

Setup and configuration

Install the package globally with npm or run it without a permanent installation. The direct no-install command is:

npx -y @signal8ai/mcp

Create an API key through Signal8’s MCP signup and API-key settings pages, then expose it as SIGNAL8_API_KEY. For Claude Desktop, add an MCP server entry with command set to npx, arguments -y and @signal8ai/mcp, and the key in the entry’s env object. Cursor uses the same command and environment-variable pattern in .cursor/mcp.json.

The README identifies Claude Desktop and Cursor configuration examples. Other MCP clients can run the same process with the environment variable set before invocation. Keep the key outside source control and replace example key text with the real credential before use.

Tools and capabilities

Representative capabilities include:

• Find companies and retrieve profiles, quotes, financials, earnings, news, analyst data, and executive information.
• Analyze short interest, float composition, historical prices, market metrics, movers, breadth, and trading halts.
• Search SEC filings, inspect filing metadata, read filing content, and list exhibits.
• Research Form 4 insider activity, 13F institutional ownership, counterparties, counsel, and cross-company patterns.
• Monitor listing deficiencies, compliance alerts, exchange classifications, dilution risk, warrants, convertibles, ATMs, and shelf registrations.
• Screen companies and filings using market, sector, industry, exchange, form, date, and other supported filters.
• Query earnings, economic, filing-deadline, clinical-trial, and material-event calendars.

Tool availability should be checked against the runtime because the README says some documented tools and categories are currently disabled.

Limitations and notes

The API key is required, so this server is not credential-free. Results are subject to Signal8’s account credit model. Large filing documents can be substantial; the filing-document tool supports maxLength to truncate previews.

Float history is forward-only from mid-2026 and is not backfilled. An empty result can therefore be normal rather than evidence that a company was delisted. Each row identifies its source, including providers such as Polygon, computed data, SEC 10-K data, or FMP, and those sources may differ in quality.

Quote responses include trading-halt fields. A halted ticker may return its last known quote, or a null current price when no recoverable quote exists. Market-wide endpoints also have normal empty states: no active trading halts and no results for a non-trading historical date are not necessarily errors. The supplied material also notes that the latest market-news endpoint may return 404 until its backend deployment is available.

The signal8ai/signal8-mcp MCP server provides data and analysis inputs; the supplied material does not establish that it offers investment advice, guaranteed real-time coverage, or complete historical backfills.