kagi extract

Extract the readable content of a web page as markdown using Kagi's Extract API.

Synopsis

kagi extract <URL> [--format markdown|json|compact]
kagi extract --filter

Description

The kagi extract command sends one HTTPS URL to Kagi's v1 Extract API and prints the extracted page markdown to stdout by default. It is useful when a search result, article, or documentation page needs full-page text instead of a search snippet.

The command uses JSON mode internally because that is the stable envelope returned by the API, then prints the first page's markdown field. Use --format json or --format compact to print the full response envelope, including retained link metadata when Kagi returns it. If Kagi returns no page markdown, the CLI reports the Extract API error details and trace id when available.

Use --filter for pipelines that already have one URL per line. Filter mode ignores the single-URL markdown default and prints one compact JSON record per input line in the same order. Successful records include the Extract API response under response. Failed records include an error envelope on stdout and the command exits nonzero after all inputs are processed.

Authentication

Required: KAGI_API_KEY

Extract is part of Kagi's current /api/v1 paid API surface, uses Bearer auth, and consumes API credit per request.

Arguments

<URL> (Required)

The HTTPS URL of the page to extract.

kagi extract "https://example.com/article"

Only https:// URLs with a valid host are accepted.

Omit <URL> when using --filter.

--filter

Read one URL per stdin line and print JSONL records.

kagi search "rust release notes" --template '{{url}}' | kagi extract --filter

--format <FORMAT>

Output format. Default: markdown.

Supported values:

• markdown - print the first page's markdown content only
• json - print the full Extract API response envelope
• compact - print the full response envelope as compact JSON

Output

The command prints markdown directly:

# Article title

Extracted page content...

With --format json, the command prints the full response envelope:

{
  "meta": {
    "trace": "trace-1",
    "node": "test",
    "ms": 12
  },
  "data": [
    {
      "url": "https://example.com/article",
      "markdown": "# Article title\n\nExtracted page content...",
      "links": [
        {
          "text": "Source",
          "url": "https://example.com/"
        }
      ]
    }
  ]
}

With --filter, stdout is JSONL:

{"url":"https://example.com/article","ok":true,"response":{"meta":{"trace":"trace-1","node":"test","ms":12},"data":[{"url":"https://example.com/article","markdown":"# Article title\n\nExtracted page content...","links":[]}]}}
{"url":"http://example.com/bad","ok":false,"error":{"code":"configuration_error","category":"configuration","retryable":false,"message":"configuration error: extract URL must use the https scheme","suggested_commands":[],"docs_url":"https://kagi.micr.dev/reference/error-reference"}}

Examples

Save an Article

kagi extract "https://example.com/article" > article.md

Pipe into Another Tool

kagi extract "https://example.com/article" | sed -n '1,80p'

Inspect Retained Links

kagi extract "https://example.com/article" --format json | jq '.data[0].links'

Extract Search Results

kagi search "rust async cancellation" --template '{{url}}' | kagi extract --filter > pages.jsonl

Exit Codes

Common errors:

• Missing API key
• Non-HTTPS or invalid URL
• Insufficient API credit
• Kagi returns no extractable content
• Network error