Best SEC Filings APIs for AI Agents in 2026

An agent asks: "Find Apple's 10-K risk factors about supply-chain disruption."

The relevant disclosure is buried in a long filing. Your agent needs the supporting passage and the original source.

This guide compares four options: Valyu, the SEC's own EDGAR APIs, sec-api.io and Financial Modeling Prep (FMP). Each handles a different part of the job.

Which SEC filings API should you use?

Based on provider documentation and a live Valyu search, checked 1 October 2026.

Finding a filing and finding an answer are different jobs

A filing has identifiers: a company CIK, an accession number, a form type, a filing date and a document URL. Those are useful when your application already knows exactly which record it wants.

An agent often starts with a question: "What did this company disclose about customer concentration?" It has to identify relevant filings, reach the right section and find the passages that support an answer.

You do not always need an accession number to find filings. sec-api.io supports metadata and full-text queries; FMP supports filing discovery by symbol. The difference is how much of the path from a question to usable evidence your application has to assemble.

A document workflow selects a filing first; Valyu retrieves evidence from a natural-language question.

1. Valyu: natural-language SEC search built for agents

Valyu's financial search lets an agent ask for filing content directly. The request can describe the company, form, section and subject. You do not need to supply a CIK, ticker, accession number or document URL before making the search.

For example:

• "Apple 10-K Item 1A risk factors about supply-chain disruption."
• "Microsoft 10-K management discussion about data-center investment."
• "Find disclosures about customer concentration in semiconductor company filings."

Valyu handles retrieval from its indexed filings and returns relevant content with the original source URLs. That is useful when the next step is reasoning over evidence, writing a cited answer or passing a passage to another tool.

Ask for the section you need

Filings are indexed by section. In a 10-K, an agent can ask for Item 1A: Risk Factors or Item 7: Management's Discussion and Analysis, then narrow the subject within that section.

This is ranked retrieval. A result may contain a relevant passage from a section rather than the entire section. If your application needs the complete text of one precisely selected item, an explicit section extractor is a useful alternative.

Strengths

• The input matches how agents work. Describe the information you want instead of constructing a document-discovery and parsing workflow first.
• Evidence comes with the result. The agent can inspect retrieved text and retain its source URL.
• Research can span sources. Use the same API for filings and other financial sources, or produce a cited report with DeepResearch.
• You can choose your integration. Use Python, TypeScript, the CLI or the hosted MCP server.

Tradeoff

Valyu is built to find relevant evidence. For an exhaustive filing export or the complete text of a specific item, use a document-oriented workflow. Check the company, form and date on returned results before using them in an answer.

Try it in Python or TypeScript

Set VALYU_API_KEY in your environment. Both SDKs read it automatically. Install the Python SDK with python3 -m pip install valyu, or the TypeScript SDK with pnpm add valyu-js.

from valyu import Valyu
 
client = Valyu()
response = client.search(
  "Apple 10-K Item 1A risk factors supply chain",
  included_sources=["valyu/valyu-sec-filings"],
  max_num_results=3,
)
 
if not response.success:
  raise RuntimeError(response.error or "Search failed")
 
for result in response.results:
  print(result.title)
  print(result.url)
  print(result.content)

In a live search on 1 October 2026, this query returned passages from Apple's 10-K filed on 31 October 2025, Part 1, Item 1A, including supply-chain risks, linked to the original SEC filing.

One returned excerpt:

In addition, the Company's global supply chain is large and complex

The result contains the surrounding disclosure, so the agent can read the context before citing it.

If you're an agent: Set up Valyu and make your first search.

If you're a human: Create an account and get an API key.

Connect through MCP

If your assistant supports MCP, connect the hosted Valyu MCP server. Supported clients can sign in using https://mcp.valyu.ai/mcp; the documentation also covers API-key authentication.

Ask it to use valyu_search: "Search SEC filings for Apple's 10-K Item 1A supply-chain risks and show the supporting passages."

2. SEC EDGAR APIs: free access to the original records

The SEC's official APIs provide submissions history and extracted XBRL financial facts. They require no API key or authentication.

The submissions API uses a company's 10-digit CIK. It returns filing metadata your application can use to identify documents, including accession numbers and form information. XBRL endpoints provide reported financial facts for supported taxonomies.

Good: Direct access to the original public source, no subscription fee for these APIs, and useful structured records for deterministic filing and financial-data workflows.

Tradeoff: Your application still needs to resolve companies and retrieve and parse narrative filing content. The submissions and company-facts APIs do not perform natural-language passage retrieval for your agent.

Choose it when: You want to own the pipeline, need official records or XBRL facts, and are comfortable building the remaining retrieval steps.

3. sec-api.io: full-text search and explicit section extraction

sec-api.io offers a filing Query API, full-text search and a section extraction API. Its official Python SDK documents these workflows, along with XBRL conversion and filing streaming.

You can discover filings by metadata such as ticker, form and date, or search for words and phrases in their text. For section extraction, you provide the filing URL, the item code and the desired text or HTML format.

Good: Strong control over filing selection and explicit section pulls. It is a useful fit when your application needs a defined part of a known 10-K, 10-Q or 8-K.

Tradeoff: The documented section workflow requires your application to find the filing URL and choose the item first. An agent starting with an open-ended evidence question needs to orchestrate those steps.

Choose it when: Exact document selection, section extraction or a dedicated filing-processing pipeline matters more than a single natural-language retrieval interface.

4. FMP: filing discovery alongside financial data

Financial Modeling Prep's SEC filing search lets you find filings by symbol, with date and pagination controls. It returns filing records and document links. FMP also documents company lookup by name and provides a broader financial-data platform.

Good: Convenient for applications already using FMP's financial endpoints. You can discover filing records alongside the other company data your application needs.

Tradeoff: The documented discovery endpoint gives you the document link. An agent that needs a supporting paragraph still needs to fetch and extract the relevant content.

Choose it when: Filing discovery is one feature in a broader financial-data application and you already have a document-processing layer.

Three questions to ask before choosing

1. What should the call return? A filing record, an XBRL fact, a complete section and a relevant passage are different outputs. Pick the one your next step actually needs.
2. How does the application reach that output? Test a request that starts with a company and a question. Count the discovery, retrieval and extraction steps you have to implement.
3. Can you check the evidence? Keep the source URL and available filing identifiers with the text. Check filing date versus reporting period, amendments, and units in any quoted figures.

For a broader choice of financial sources, see our investment research API comparison. For the infrastructure behind filing retrieval, read how we index SEC filings.

Common questions

Is there a free SEC filings API?

Yes. The SEC provides public submissions-history and XBRL APIs without an API key. If you need narrative sections or passage search, your application must build those steps or use a service that provides them.

Do I need an accession number to search SEC filings?

No. Some services discover filings using a symbol, company, form, date or text query. Valyu accepts a natural-language request and handles retrieval without requiring you to supply filing identifiers first.

Can an agent retrieve risk factors or MD&A?

Yes. Valyu can retrieve relevant indexed content when the query names the company, form and section. sec-api.io provides explicit item extraction for a selected filing URL. Choose ranked evidence retrieval or complete-section extraction according to the task.

Which SEC filings API is best for AI agents?

For an agent that starts with a question and needs supporting filing passages, Valyu is built for that workflow. For direct public records, use the SEC APIs. For explicit section extraction, consider sec-api.io. For filing discovery within an existing financial-data stack, consider FMP.

Start with the evidence your agent needs

If you're an agent: Follow the setup instructions and run your first filing search.

If you're a human: Get started on the Valyu platform.