mcpField
F
FilingLens Local SEC Fact Comparison

FilingLens Local SEC Fact Comparison

Local
filinglens.agiscorecard.comFinance0Updated today

About FilingLens Local SEC Fact Comparison

FilingLens is a local, read-only MCP server for comparing an SEC financial fact at two filing cutoffs for the same period. It is intended for financial filing review and reports source references alongside exact comparisons.

Key features

  1. 01Compare facts across filing cutoffs
  2. 02List exact reporting periods
  3. 03Use supplied SEC companyconcept data
  4. 04Preserve source references
  5. 05Handle decimal differences with integer arithmetic

Use cases

  1. 01Review a reported fact across two filing dates
  2. 02Check available period and unit combinations
  3. 03Trace a comparison to its filing references

MCP support

It provides three read-only tools, a review prompt, and a contract resource to MCP clients. The host supplies an official SEC companyconcept JSON object; the package makes no network requests.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

No install method listed yet

We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.

Repository
Packages
mcpbfilinglens-0.1.0.mcpb

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

FilingLens MCP — internal evaluation beta

Compare the same SEC financial fact for the exact same period at two filing cutoffs. Three read-only tools: filinglens_example, filinglens_periods, filinglens_compare. Prompt: review_filing_change; resource: filinglens://contract.

Requires Node.js 22+. Extract the published archive, verify its SHA-256, then run npm ci --ignore-scripts --no-audit --no-fund and npm test. Run node scripts/evaluate.mjs for the ten independent protocol scenarios. Configure an MCP host with command node and absolute argument /path/filinglens-mcp/src/server.mjs. A plain terminal waits on stdio; it is not a chat interface.

The host supplies one official SEC companyconcept JSON object. No file reads, network, storage, model calls or telemetry occur in this package. Host fees and data policies apply. The browser site has a separate, cached official SEC lookup. Source documentation: https://www.sec.gov/search-filings/edgar-application-programming-interfaces and https://www.sec.gov/search-filings/edgar-search-assistance/accessing-edgar-data . Respect SEC fair access rules when obtaining data.

filinglens_periods returns exact start/end/unit combinations, with pagination. filinglens_compare accepts {data, options:{unit,start:null|"YYYY-MM-DD",end,before,after}}. before must precede after. It picks the latest eligible filing date at each cutoff; same-day value conflicts are incomplete. A missing baseline is not zero. It never combines fiscal labels, quarter frames, concept tags or units. Absolute decimal differences use integer arithmetic; unsafe numbers are rejected. Source references are supplied metadata and must be checked in original filings. Fictional example records have no source links.

Supported concepts: Assets, Liabilities, StockholdersEquity, Revenues, RevenueFromContractWithCustomerExcludingAssessedTax, NetIncomeLoss, EarningsPerShareDiluted, NetCashProvidedByUsedInOperatingActivities. Custom dimensions and tags are outside scope. No restatement classification, alerts, trading or investment recommendation. A reported-value change does not prove an accounting error, its cause or an investment opportunity.

The free beta license permits internal evaluation only. The proposed €19/month workspace is not built or for sale. No revenue or customer proof is claimed. Validation includes adversarial fixtures and a real SDK stdio client, not a benchmark of host-model reasoning or every desktop host.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.