How do I install the Fetch Text skill?

Run npx skillmds@latest add bdambrosio/fetch-text in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Fetch Text safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Fetch Text?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Fetch Text free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Fetch Text?

bdambrosio (@bdambrosio) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Fetch Text by bdambrosio — Agent Skill

Name: Fetch Text
Author: bdambrosio

File contents

fetch-text

Retrieve and extract the text of a URL. Auto-detects the format (HTML / PDF / Markdown / plain text).

Behavior

Research PDF (parsed into sections): the first call returns the title, the abstract, and a section index — every section name with its size — rather than the body text. Call again with the same url plus section to read one section. A long paper runs to tens of thousands of words; the index lets you read the two sections that answer the question instead of flooding the conversation with the rest.
Other PDFs: page-by-page text extraction, truncated at 8000 chars.
HTML: structured text with links and image references preserved.
Markdown / plain: returned as-is.
Page candidate images (Open Graph, Twitter cards, in-page <img> tags) are surfaced as part of the result so a downstream display can pick a primary image without re-fetching.

Reading a paper

{"tool": "fetch-text", "url": "<pdf url>"} → title, abstract, section index.
Pick the sections that bear on the question — not all of them.
{"tool": "fetch-text", "url": "<same url>", "section": "Results"} → that section.

Asking for a section that isn't in the index returns an error listing the names that are, so re-read the index rather than guessing. Section names match case-insensitively, and an unambiguous prefix works.

For a paper's reference list, use semantic-scholar with a paper_id — the citation graph gives resolved records, which beats parsing a bibliography out of the PDF.

When to use vs search-web

fetch-text — you already have a specific URL and want its content.
search-web — you're looking for information and don't yet have the URL.

Examples

{"thought": "see what's in this paper", "tool": "fetch-text", "url": "https://arxiv.org/pdf/1706.03762"}

{"thought": "the index listed a Results section — read it", "tool": "fetch-text", "url": "https://arxiv.org/pdf/1706.03762", "section": "Results"}

{"thought": "load today's weather forecast page", "tool": "fetch-text", "url": "https://forecast.weather.gov/MapClick.php?lat=37.87&lon=-122.27"}

Notes

Section support needs a reachable GROBID server. GROBID_URL overrides the default http://localhost:8070/api/processFulltextDocument; set it to an empty string to disable. With GROBID unavailable, PDFs fall back to flat page text and the section arg has nothing to resolve against.

---name: fetch-textdescription: Fetch the text content of a web page or PDF from a URL. Handles HTML, PDF, Markdown, plain text — auto-detects format. Research PDFs come back as a section index you can then read section by section.---# fetch-textRetrieve and extract the text of a URL. Auto-detects the format (HTML /PDF / Markdown / plain text).## Behavior- **Research PDF** (parsed into sections): the first call returns the title, the abstract, and a **section index** — every section name with its size — rather than the body text. Call again with the same `url` plus `section` to read one section. A long paper runs to tens of thousands of words; the index lets you read the two sections that answer the question instead of flooding the conversation with the rest.- **Other PDFs**: page-by-page text extraction, truncated at 8000 chars.- HTML: structured text with links and image references preserved.- Markdown / plain: returned as-is.- Page candidate images (Open Graph, Twitter cards, in-page `<img>` tags) are surfaced as part of the result so a downstream `display` can pick a primary image without re-fetching.## Reading a paper1. `{"tool": "fetch-text", "url": "<pdf url>"}` → title, abstract, section index.2. Pick the sections that bear on the question — not all of them.3. `{"tool": "fetch-text", "url": "<same url>", "section": "Results"}` → that section.Asking for a section that isn't in the index returns an error listing thenames that are, so re-read the index rather than guessing. Section namesmatch case-insensitively, and an unambiguous prefix works.For a paper's **reference list**, use `semantic-scholar` with a`paper_id` — the citation graph gives resolved records, which beatsparsing a bibliography out of the PDF.## When to use vs search-web- `fetch-text` — you already have a specific URL and want its content.- `search-web` — you're looking for information and don't yet have the URL.## Examples```json{"thought": "see what's in this paper", "tool": "fetch-text", "url": "https://arxiv.org/pdf/1706.03762"}``````json{"thought": "the index listed a Results section — read it", "tool": "fetch-text", "url": "https://arxiv.org/pdf/1706.03762", "section": "Results"}``````json{"thought": "load today's weather forecast page", "tool": "fetch-text", "url": "https://forecast.weather.gov/MapClick.php?lat=37.87&lon=-122.27"}```## Notes- Section support needs a reachable GROBID server. `GROBID_URL` overrides the default `http://localhost:8070/api/processFulltextDocument`; set it to an empty string to disable. With GROBID unavailable, PDFs fall back to flat page text and the `section` arg has nothing to resolve against.

Frequently asked questions

Fetch Text