page-reader
Public
README.md
README.md
Give your model a link and it reads the page. An article, a documentation page, a PDF report: it comes back as clean text without the menus and clutter, so the model can summarise it, quote it or answer questions from what it actually says.
Project home · Report a problem
"Summarise https://example.com/blog/some-post in five bullet points."
"Read this PDF and tell me what it says about delivery times: https://example.com/report.pdf"
"What does the pricing table on https://example.com/pricing say? Only the table."
| Tool | What it does |
|---|---|
fetch_url | Downloads a web page and returns its main content as markdown. |
Pages that need JavaScript come back nearly empty. This plugin downloads the page as the site sends it and does not run scripts, so a site that builds its content in the browser has little or nothing to read. The tool says so when that happens. For those sites you need a real browser, such as the browser tools in agent-toolkit.
| Setting | Default | |
|---|---|---|
| Max Characters | 15000 | How much of a page comes back at once. |
http and links are fetched. Anything else, such as a link to something on your computer, is refused, and so are links with a username or password in them.Using agent-toolkit? Its Web group already includes fetch_url, so you don't need this plugin as well. Enabling both in one chat gives the model the same tool twice.
MIT licensed · Part of the LM Studio Agent Toolkit
Give your model a link and it reads the page. An article, a documentation page, a PDF report: it comes back as clean text without the menus and clutter, so the model can summarise it, quote it or answer questions from what it actually says.
Project home · Report a problem
"Summarise https://example.com/blog/some-post in five bullet points."
"Read this PDF and tell me what it says about delivery times: https://example.com/report.pdf"
"What does the pricing table on https://example.com/pricing say? Only the table."
| Tool | What it does |
|---|---|
fetch_url | Downloads a web page and returns its main content as markdown. |
Pages that need JavaScript come back nearly empty. This plugin downloads the page as the site sends it and does not run scripts, so a site that builds its content in the browser has little or nothing to read. The tool says so when that happens. For those sites you need a real browser, such as the browser tools in agent-toolkit.
| Setting | Default | |
|---|---|---|
| Max Characters | 15000 | How much of a page comes back at once. |
http and links are fetched. Anything else, such as a link to something on your computer, is refused, and so are links with a username or password in them.Using agent-toolkit? Its Web group already includes fetch_url, so you don't need this plugin as well. Enabling both in one chat gives the model the same tool twice.
MIT licensed · Part of the LM Studio Agent Toolkit
selector, a CSS selector such as main or table#prices, only the matching elements come back: much shorter than the whole page.refresh fetches a fresh copy.httpsfile:selector, a CSS selector such as main or table#prices, only the matching elements come back: much shorter than the whole page.refresh fetches a fresh copy.httpsfile: