MCP Public Company Reporting Window Finder

What the mcp-public-company-reporting-window-finder MCP server does

The mcp-public-company-reporting-window-finder MCP server exposes five modes for working with public-company identity, classification, reporting timing, and reporting-season data. It runs the corresponding Mamba Labs actor on Apify and returns structured results rather than requiring a single multipurpose mode.

Company-oriented lookups can start with a bare domain, legal or trading name, ticker, ISIN, LEI, or SEC CIK. The identity mode returns company and venue information, while the qualification mode produces a listed-status verdict. The timing mode adds fiscal year end, reporting cadence, an expected next event, days to that event, and a configurable outreach window.

Filter-based modes serve different purposes. One creates a company universe, and the other groups expected reporting load by week or month. These modes use filters instead of company identifiers and can constrain results by geography, venue, sector, security type, public-float band, cadence, fiscal year end, issuer status, and data-quality conditions.

How it works

Each tool sets its own actor mode; callers do not pass a separate mode parameter. Identifier-based tools accept one or more supported identifiers, while universe and season tools select rows through filters. Exchange codes use ISO 10383 MICs, country filters use ISO 3166-1 alpha-2 codes, and regions provide shorthand for groups such as the US, UK, and EU.

Reporting windows are calculated from a selected event type and lead and lag values. The default window opens 70 days before an event and closes 42 days before it. Timing results can be restricted by window state, maximum days to the event, cadence confidence, and event type. Season results can be grouped by week or month and optionally split by sector, country, or exchange.

Setup and configuration

Configure the server as an MCP stdio process using npx and provide an Apify token through APIFY_TOKEN:

{
  "mcpServers": {
    "mamba-public-company-reporting-window-finder": {
      "command": "npx",
      "args": ["-y", "@mambalabsdev/mcp-public-company-reporting-window-finder"],
      "env": { "APIFY_TOKEN": "your-apify-token" }
    }
  }
}

The token is obtained from the Apify account integrations page. Set limit explicitly for universe and season requests: the default is 1000, and returned rows are charged per row. Truncation is reported rather than hidden.

Tools and capabilities

• resolve_company resolves identifiers into company identity and venue data.
• qualify_company returns a listed-status verdict, with options to keep listed companies, suppress proven listed companies, or interpret unmatched companies as private.
• get_reporting_timing returns fiscal year end, cadence, next event, confidence information, and outreach-window fields.
• build_company_universe returns companies matching the supplied filters.
• get_reporting_season summarizes reporting load over a date range by week or month.

The mcp-public-company-reporting-window-finder MCP server supports filters for operating-company status, blank checks, foreign private issuer status, SEC registration, provenance confidence, name-only identity matches, and share-alike-derived sources. Timing can also include the period in which a listed company is constrained in what it can announce.

Limitations and notes

Reporting dates are estimates derived from filing history, not confirmed announcements. Timing rows cover US companies only. Identity, venue, and classification can resolve worldwide across the publishable universe, but a non-US company returns a stated timing-unavailable reason instead of an inferred date.

A timing result can be withheld when cadence confidence is below the requested threshold. The output includes window states such as open, not_yet, closed_passed, and no_event; use this field when filtering outreach opportunities. Sector data is populated for roughly 65 percent of the publishable universe, and fiscal-year-end information is effectively a US field in this dataset. The default timing confidence differs by cadence, with quarterly results generally better supported than annual or semiannual timing.