Extract Filing File

Ingest a trusted local markdown filing and run a selected extraction schema.

Tier: internal | Cache: llm-call | Playground: authenticated only | Contract: 1.0.0

When To Use It

Use these tools for curated extraction caches and premium language-model-backed filing evidence.

This is an MCP-only tool. Call it through an MCP client after configuring edgar-mcp.


{
  "tool": "extract_filing_file",
  "arguments": {
    "file_path": "/tmp/filing.html",
    "schema_name": "operational_kpis"
  }
}

schema_name, evidence spans, and source filing metadata for every extracted claim.
Coverage or cache-miss fields before treating missing rows as “not disclosed.”
Confidence, provenance, and extraction timestamp fields when comparing periods.

401 or 403: the API key is missing, expired, or not allowed to use this tier.
429: retry after the rate-limit window, and prefer broader batch or series tools over repeated single-period calls.
Paid or LLM-backed paths should be used only when cheaper fact, filing, table, or text tools cannot answer the question.
Treat partial coverage, empty arrays, and warning fields as answer-quality signals even when HTTP status is 200.