lb capture

Synopsis

lb capture [url] [options]

Description

lb capture adds content to your knowledge graph. It accepts a URL to fetch and process, a local file to upload, a short inline note, or content piped from stdin.

Input methods

Inputs are resolved in the following priority order. Only one source is used per invocation.

Priority	Source	How to trigger
1	File upload	--file <path>
2	URL	Positional argument
3	Inline note	--note <text>
4	Stdin pipe	Pipe content to lb capture

Options

Option	Type	Description
--why <text>	string	Required. Why you are saving this. Used for ranking and recall.
--note <text>	string	Short inline note. Used as content when no URL or file is provided.
--title <text>	string	Override the auto-detected title.
--type <type>	string	Content type hint. See content types below.
--file <path>	string	Path to a local file to upload. Max 25 MB.
--json	flag	Output raw JSON.
--api-url <url>	string	Override the API base URL.

Content types

Type	Description
article	Web article or blog post (default for URLs)
tweet	Twitter/X post
repo	GitHub or similar code repository
pdf	PDF document
audio	Audio file
note	Plain-text note (default for --note and stdin)
video	Video content
image	Image file

When --type is omitted, the CLI infers the type from the URL domain or file extension.

Limits

Input	Limit
File upload (--file)	25 MB
Stdin pipe	5 MB

Examples

Capture a URL:

lb capture https://example.com/article --why "reference for the auth redesign"

Capture a quick note:

lb capture --note "Check the Drizzle docs for batch insert syntax" --why "ongoing reference"

Pipe content from stdin:

curl -s https://example.com/raw.md | lb capture --why "downloaded spec"

Upload a local PDF:

lb capture --file ./report.pdf --why "Q4 analysis" --type pdf

Override the title and type:

lb capture https://github.com/org/repo --why "dependency audit" --type repo --title "Org Monorepo"

Output as JSON for scripting:

lb capture https://example.com --why "research" --json | jq '.id'