> ## Documentation Index
> Fetch the complete documentation index at: https://docs.secapi.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Install SEC API MCP

> Connect an MCP client to SEC API, authenticate with an API key, and verify a filing workflow

Connect an MCP client to SEC API when you want filing, company, and research tools available inside that client. Authenticate once, then confirm that the client can return a specific SEC filing before using it in a larger workflow.

## Set the API key

Create an API key in the signed-in [dashboard](https://secapi.ai/app), then expose it only to the MCP client process:

```bash theme={null}
export SECAPI_API_KEY="secapi_..."
```

SEC API serves MCP at `https://api.secapi.ai/mcp`. `GET /mcp` returns public connection metadata. MCP requests use JSON-RPC over an authenticated `POST /mcp` request with the key in `x-api-key`.

## Add the server

For Claude Code, add the hosted server with the API key header:

```bash theme={null}
claude mcp add --transport http secapi https://api.secapi.ai/mcp \
  --header "x-api-key: $SECAPI_API_KEY"
```

For another MCP client, use its HTTP-server configuration with the same URL and header. A typical configuration looks like this:

```json theme={null}
{
  "mcpServers": {
    "secapi": {
      "url": "https://api.secapi.ai/mcp",
      "headers": {
        "x-api-key": "${SECAPI_API_KEY}"
      }
    }
  }
}
```

Use the environment-variable syntax required by your client. Do not put an API key directly in a checked-in configuration file.

## Verify a filing workflow

Restart or reload the client after adding the server. Then give it a bounded request that requires source fields in the answer:

<Prompt>
  Resolve AAPL, retrieve its latest 10-K, and return Item 1A. Include the accession number, filing date, filing URL, and request ID from the SEC API result.
</Prompt>

A successful result identifies one filing and gives you fields to retain with any downstream analysis. Filing details change as new reports arrive, so do not compare the result to a fixed accession number.

## Discover tools before calling them

The server exposes its current tool catalog through the MCP protocol. Ask the client to list tools, or use `tools.search` to find a capability and `tools.describe` to inspect its input schema before calling it. This avoids relying on an old local catalog or a copied prompt.

## Troubleshoot a connection

If the client cannot list tools, confirm the URL, reload the configuration, and verify that `SECAPI_API_KEY` is present in the process environment. If a tool call fails, retain the HTTP status or JSON-RPC error, `requestId`, and any trace context before retrying.

Read [MCP workflows](/mcp-workflows) for task patterns and [Troubleshooting](/troubleshooting) for error-specific guidance.