Read Web

Fetch any web page as clean, LLM-friendly markdown using the Jina Reader API.

Usage

To read a web page, make an HTTP request to the Jina Reader API by prefixing the target URL with https://r.jina.ai/:

curl -s -H "Accept: text/markdown" \
     "https://r.jina.ai/https://example.com"

Prefer the unauthenticated request above first. Only add the configured bearer token if an unauthenticated request is rate-limited or the page requires it:

curl -s -H "Authorization: Bearer $(cat ~/dotfiles/scriptswithsecrets/jina)" \
     -H "Accept: text/markdown" \
     "https://r.jina.ai/https://example.com"

Options

Return format

By default, returns markdown. You can request other formats:

• Markdown (default): -H "Accept: text/markdown"
• Plain text: -H "Accept: text/plain"
• HTML: -H "Accept: text/html"

Additional headers

• No cache (force fresh fetch): -H "X-No-Cache: true"
• With links preserved: -H "X-With-Links: true"
• With images preserved: -H "X-With-Images: true"
• Target selector (extract specific element): -H "X-Target-Selector: article"
• Wait for selector (dynamic pages): -H "X-Wait-For-Selector: .content"
• Timeout (ms): -H "X-Timeout: 10000"

Examples

Read a documentation page

curl -s -H "Accept: text/markdown" \
     "https://r.jina.ai/https://docs.example.com/getting-started"

Extract just the article content

curl -s -H "Accept: text/markdown" \
     -H "X-Target-Selector: article" \
     "https://r.jina.ai/https://blog.example.com/post"

Read a page with links preserved

curl -s -H "Accept: text/markdown" \
     -H "X-With-Links: true" \
     "https://r.jina.ai/https://example.com/resources"

Notes

• The API converts JavaScript-rendered pages to clean text, so it works on SPAs and dynamic sites.
• For very large pages, consider using X-Target-Selector to extract only the relevant section.
• If an authenticated request returns InsufficientBalanceError, retry without the Authorization header.
• If X-Target-Selector returns no content, retry without the selector. Some documentation sites, including Kubernetes docs pages, may not expose content under a plain article selector to Reader.