MCP Tools

edgar.tools exposes MCP tools for SEC filing intelligence, organized into eight categories: Analyze, Research, Compare, Notes, Signals, Monitor, Funds & Advisers, and Aggregation. The tools/list endpoint advertises only the tools your account's tier includes. See MCP Tiers for the full mapping, or the AI plugin landing page for the visual ladder.

Start with search_companies to resolve a company, then use the identifier with other tools. For Analyst-tier multi-tool workflows (theses, peer comparisons, portfolio pulses), see MCP Prompts — five recipe prompts that orchestrate the aggregation tools below into structured outputs.

The "Required tier" column below is the minimum tier that can call the tool — tools/list doesn't advertise it to lower tiers. Within tiers that can call a tool, depth differences still apply (a Free user calling company_brief gets a truncated profile compared to a Pro user). Per-tool depth differences are noted in each tool's section.

Analyze#

analyze_filing#

Paste any SEC filing URL or accession number. Auto-detects form type and returns structured intelligence. For prospectus filings (424B), returns syndicate details, dilution analysis, and offering lifecycle data.

Example: "Analyze this filing: https://www.sec.gov/Archives/edgar/data/320193/000032019324000081/0000320193-24-000081-index.htm"

Tier. Analyst and above. Returns full structured analysis including prospectus syndicate, dilution, and lifecycle details. Form-type aware: Form 4s get insider transactions; 8-Ks get material events; 10-Ks get financial summaries plus notes TOC; 13F-HRs get top holdings; 424Bs get the prospectus extract.

filing_details#

Progressive-disclosure filing reader — the open→drill loop of analyze_filing plus filing_section in one tool. Open a filing (pass url) to get its metadata, form-specific intelligence, a bounded digest, and an available_parts manifest of part handles with token-size hints. Fetch depth by adding part from that manifest to pull just that section's full content.

Example: "Open this 8-K, then read its press-release exhibit: 0000320193-24-000123"

Tier. Analyst and above. Open first, read the manifest, pull only the parts the question needs.

filing_section#

Return the body of a single named section of a filing as plain text (markdown for earnings_release). Use this when you need filing prose — an MD&A paragraph, a specific risk-factor body, a press release — instead of fanning the whole 10-K through the agent.

Example: "Show me the risk factors section from Apple's latest 10-K"

Tier. Pro and above. Free tier returns a 200-character preview plus an upgrade hint.

Research#

search_companies#

Search for SEC-registered companies by name, ticker, or CIK. Use this first to resolve a company before calling other tools. Free tier, no authentication required.

Example: "Find me the CIK for Tesla"

search_entities#

Federated search across every SEC entity directory — companies, funds, advisers, and persons — in one call. Returns hits[] tagged by kind, each with per-kind rich fields and recommended follow-up tools. Use search_companies instead when the query is known to be a company.

Example: "Search SEC for anything named Renaissance Technologies"

Tier. Available to all tiers. Free tier resolves company hits and the federated fan-out; Pro widens the fan-out to fund, adviser, and person hits.

company_brief#

Full intelligence snapshot for a company: profile, financials, insider sentiment, material events, and institutional holders in a single call.

Example: "Give me a full brief on NVIDIA"

Tier. Available to all tiers. Free tier returns basic profile + summary data; Pro and above return full cross-dataset analysis with financials, insider sentiment, events, and holders.

account_dossier#

Sales-shaped account dossier for a public company in one call: identity, financial snapshot, recent 8-Ks (with deterministic routing tags), current leadership plus recent officer changes, insider buy/sell signals, peer companies, and data-quality indicators. Built for sales-call prep and account research — every executive and peer carries a CIK and every event an accession, so follow-up questions chain straight into position_intel, analyze_filing, or compare_companies.

Example: "Prep me for my call with Salesforce"

Tier. Analyst and above (not advertised to lower tiers — company_brief covers the lighter "describe X" intent). Returns all sections, prior-role history, and a 365-day event window.

company_filings#

Get recent SEC filings for a company. Returns filing metadata including form type, date, and accession number.

Example: "What has Apple filed in the last 30 days?"

Tier. Available to all tiers. Free tier capped at 10 filings per call with no form filter; Pro and above unlock 100-result calls and form-type filtering.

account_status#

Get the calling account's current tier, the list of tools and prompts it can call, remaining monthly quota, and the upgrade URL. Useful when an AI agent encounters a tier_denied or quota_exceeded envelope and needs to explain the situation to the user.

Example: "Why did that last tool call fail? Check my plan."

Tier. Available to all tiers. Returns the tier-specific tool list and quota state.

edgar_recipes#

Fetch or discover a prompt-framework recipe. Recipes are LLM prompt templates (read / track / research buckets) that orchestrate edgar.tools MCP tools into analyst-quality output grounded in real filings. All arguments are optional: name returns a recipe's full markdown; query keyword-searches the library and returns a shortlist; bucket filters the list; no arguments returns the full catalog.

Example: "Show me the edgar.tools recipe for a bear-case teardown"

Tier. Available to all tiers. Browse the recipe hub at /mcp/recipes.

financial_snapshot#

Get financial ratios for a company from its latest annual filing (10-K). Covers profitability, returns, liquidity, and leverage.

Example: "What are Apple's financial ratios? Is the company healthy?"

Tier. Pro and above. Returns all 12 ratios across 4 categories (profitability, returns, liquidity, leverage) plus a health score.

financial_statements#

Get full XBRL financial statements for a company. Returns line-item detail for income statement, balance sheet, and/or cash flow statement.

Example: "Show me Tesla's full income statement from their latest 10-K"

Tier. Pro and above.

financial_trends#

Multi-year time series for financial concepts with growth rates, CAGR, and trend classification.

Example: "Show me Apple's revenue and net income trends over the last 5 years"

Tier. Pro and above. Up to 10 years of history, all concepts, trend classification and anomaly flags.

Compare#

compare_companies#

Side-by-side comparison of multiple companies with intelligence signals. Compare financials, health scores, insider sentiment, and material events.

Example: "Compare Apple, Microsoft, and Google on profitability and leverage"

Tier. Pro and above. Up to 10 companies, all metrics, health + insider + event signals.

Notes#

edgar_notes#

Drill into the financial statement notes of a 10-K or 10-Q by topic — revenue, leases, segments, debt, fair value, and more.

Example: "Show me Apple's segment reporting note from their latest 10-K"

Tier. Pro and above. Returns matched notes with their full plain-text bodies and a concatenated focused context; no matches falls back to the full notes table of contents.

disclosure_search#

Search XBRL narrative text disclosures across a company's SEC filings. Find revenue recognition policies, risk factors, segment details, and other text-block content.

Example: "What is Apple's revenue recognition policy? Show me the relevant disclosure."

Tier. Pro and above. Returns up to 25 results with full text, 400-character snippets, and form/year filters.

search_filings#

Renamed from search_filings_full_text (the old name still works as a deprecated alias until 2026-08-15 — switch to search_filings).

Full-text search across every SEC filing since 2001. Supports phrase quotes ("material weakness"), boolean OR/NOT, NEAR proximity (NEAR(revenue, decline, 5)), and suffix wildcards (gas*). Returns matched filings with highlighted snippets from the document body — or, in trend mode, a count-over-time series instead of hits.

When both date_from and date_to are omitted, the search defaults to the last 12 months (the response reports this as applied_default_window) — pass date_from: "2001-01-01" to search all of EDGAR history.

A plain query is widened into the filing-language phrases it maps to — buyback searches "share repurchase program" too — so the topic isn't missed just because the filer used formal wording. The response echoes what was added in also_matched; pass expand: false for a literal-only search. This applies in trend mode too: the whole series then counts the expanded phrase set.

Example (search): "Find every 10-K filed since 2024 that mentions material weakness within 5 words of internal control"

Example (trend): "How many 8-Ks mentioned 'restructuring' each month for the last 12 months?" → interval: "month", periods: 12 returns series[] { start, end, count }. Counts are document counts (a filing matched in N documents counts N times).

Tier. Pro and above. Pro: 50 per call, paginate up to 500. Analyst: 50 per call, paginate up to 1,000. Enterprise: paginate up to the EFTS hard ceiling (10,000).

Signals#

insider_activity#

Get insider trading activity (Form 4) for a company. Shows who is buying, selling, or exercising options with transaction details and overall sentiment analysis.

Example: "Show me insider trades at Apple in the last 60 days"

Tier. Pro and above. Full transaction history with purchase/sale filtering and analysis hints.

institutional_ownership#

Get 13F institutional holders for a company. Shows which institutions hold the stock, position sizes, and changes quarter over quarter.

Example: "Who are the largest institutional holders of Tesla?"

Tier. Pro and above. Full holder list with quarter-over-quarter history.

material_events#

Get material events (8-K filings) for a company with item classification. Shows M&A activity, leadership changes, earnings releases, and other significant corporate events.

Example: "What material events has Microsoft reported recently?"

Tier. Pro and above. Full event history with item code filtering and high-impact hints.

Monitor#

live_filings#

Get recent SEC filings from a real-time feed. Without filters, returns a summary of activity with notable 8-K highlights. With filters (form type, company), returns individual filings.

Example: "What SEC filings were submitted in the last hour?"

Tier. Pro and above. Full filing details with filtering by form type and company; 48-hour window.

Funds#

search_funds#

Search mutual funds and ETFs by name.

Example: "Find Vanguard S&P 500 funds"

Tier. Pro and above.

fund_profile#

Get fund details including holdings, expense ratios, and returns.

Example: "What are the top holdings and expense ratio for VOO?"

Tier. Pro and above. Up to 25 holdings, per-class expenses, and returns history.

Advisers#

search_advisers#

Search SEC-registered investment advisers by name, with optional state filtering.

Example: "Find investment advisers named Bridgewater"

Tier. Pro and above.

adviser_profile#

Get detailed information about a specific SEC-registered investment adviser by CRD number. Includes AUM, client types, fee structures, and compliance details.

Example: "Show me the profile for Bridgewater Associates"

Tier. Pro and above. Full profile with executives, private funds, related persons, and compliance details.

Aggregation#

Aggregation tools return structured, multi-signal envelopes in one tool call. position_intel, peer_facts, and portfolio_events are designed to be consumed by the recipe prompts — a recipe synthesizes them into a thesis, peer table, or portfolio summary; person_intel bundles one person's full SEC footprint. All are Analyst-tier.

position_intel#

Single-call intelligence bundle for one ticker. Returns recent catalysts (8-K filings within the lookback), ownership shifts (insider Form 4 clusters + 13F net flow), and the risk-factor diff between the two latest 10-Ks.

Example: "Pull a position intel bundle for NVDA over the last 90 days"

Tier. Analyst and above. Pairs with the `position_thesis` recipe prompt to compose a structured single-name thesis from one tool call.

peer_facts#

Structured peer-comparison rows for a target ticker against up to 10 peers along one dimension. Returns aligned rows so a recipe can drop them into a comparison table without per-peer post-processing.

Example: "Run peer_facts on ARCC against MAIN, BXSL, PSEC, OBDC on financial_snapshot"

Tier. Analyst and above. Pairs with the `peer_comparison` recipe prompt for narrative synthesis.

portfolio_events#

Bulk 8-K material events across a portfolio of up to 50 tickers. Returns per-ticker structured events so a recipe prompt can synthesize themes, rank the most material, and call out quiet names.

Example: "Pull portfolio_events for [MAIN, ARCC, BXSL, PSEC] since 2026-04-01"

Tier. Analyst and above. Pairs with the `portfolio_pulse` recipe prompt for cross-portfolio narratives.

person_intel#

One person's full SEC footprint by CIK or name (e.g. Carl Icahn, Bill Ackman) — for activist tracking, insider/officer watch, and D&O / KYC background. Returns the person's Form 3/4/5 insider roles, Schedule 13D/G activist positions (with intent), Form ADV control roles, and trade history in one call. Use this when the subject is a person; use insider_activity for a company's Form 4 insiders — activists who file 13D personally never appear there.

Example: "What has Carl Icahn filed across the SEC?"

Tier. Analyst and above.

Common 8-K item codes#

Use these codes with the item parameter on material_events.

To summarize#

• Start with search_companies to find a company, then use the identifier with other tools.
• The tools/list endpoint advertises only your tier's tools. See MCP Tiers.
• All tools are read-only — they query SEC data, never modify anything.
• Claude automatically picks the right tool based on your question.
• For multi-tool workflows — single-name theses, peer comparisons, portfolio pulses, earnings postmortems, and red-flag scans — see MCP Prompts for the five recipe prompts that orchestrate the aggregation tools above.