agentleFS
DevinoSolutionsDevinoSolutions/claude-plugins/skills/reading-web-pages/SKILL.md

reading-web-pages

Skill0 starsChanged 9 days agoMITView on GitHub

Reads a public web page through upAPI and turns it into clean Markdown Claude can quote, summarize, or compare: article or full-page mode, links and images on or off, length capped. Use when the user asks 'read this article', 'summarize this page', 'turn this URL into Markdown', 'what does this docs page say', 'pull the content from these three links', or pastes a URL and wants its text. Read-only; spends upAPI quota.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 1 is distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Every tool in this skill comes from the upapi MCP server, the upAPI connector, and is written here as upapi:<tool>.

    line 13
  2. Very specific:

    Tools listed by name (the plugin's ?tools=claude surface): call them directly, for example upapi:fetch_markdown_post.

    line 15
  3. Very specific:

    Call upapi:fetch_markdown_post.

    line 29
  4. Very specific:

    Check the result: if finalUrl differs from the request, say where the page redirected.

    line 30
  5. Very specific:

    RATE_LIMITED: wait the number of seconds the message gives before one retry.

    line 37
  6. Distinctive:

    Do not fetch pages behind the user's own login, or pages they did not ask for.

    line 43

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md
<!-- Generated from plugins/upapi/skills/reading-web-pages/SKILL.md by scripts/sync-skills.mjs. Edit the plugin copy, then run the script. -->

upAPI: reading web pages

Fetch a page the user names and work from its real content, not from memory.

Connecting and calling the tools

Every tool in this skill comes from the upapi MCP server, the upAPI connector, and is written here as upapi:<tool>.

  • Tools listed by name (the plugin's ?tools=claude surface): call them directly, for example upapi:fetch_markdown_post.
  • Only upapi:search_ops and upapi:call_op listed (the plain URL's compact table): call upapi:call_op with the operation's slug and its input, for example { "slug": "fetch-markdown.post", "input": { "url": "https://example.com" } }. Same account, quota and errors.
  • No upapi tools at all: upAPI is not connected yet. Follow the setup-upapi skill.

A failed operation comes back as an error result whose text starts with its code, for example INVALID_INPUT: or QUOTA_EXCEEDED:.

Tools you will use

  • upapi:fetch_markdown_post (slug fetch-markdown.post, 3 units per call): fetches the page over plain HTTP and converts it to Markdown. Input: url (required, absolute http or https), mode (article, the default, extracts the main content and falls back to the whole page if too much was stripped; full keeps the whole body without navigation, footer and forms), includeLinks (default true), includeImages (default false), maxChars (500 to 400000, default 100000), timeoutMs. Returns markdown, title, byline, lang, excerpt, links, finalUrl and httpStatus, and sets truncated when it cut the text.

Workflow

  1. Take the URL from the user. If they gave none, ask; never guess an address.
  2. Pick the mode: article for a blog post, news story or docs page; full when the user wants everything on the page (a pricing table, a list page) or article came back too short.
  3. Call upapi:fetch_markdown_post. For several URLs, call it once per URL; each call costs 3 units, so say how many pages you are about to read when the user gives a long list.
  4. Check the result: if finalUrl differs from the request, say where the page redirected. If truncated is set, say so and offer a larger maxChars.
  5. Answer from the Markdown: summarize, quote, or compare as asked, and cite the page by its title and finalUrl.

When it fails

  • A client-rendered page: the tool uses no browser, so a page that builds its content in JavaScript comes back as an error, not as empty content. Offer the capturing-web-pages skill for a screenshot instead.
  • INVALID_INPUT: private, loopback and link-local addresses are refused, and the URL must be absolute. Fix the input; do not retry the same call.
  • RATE_LIMITED: wait the number of seconds the message gives before one retry.
  • QUOTA_EXCEEDED: the workspace's monthly quota is spent. Tell the user; do not retry.

Rules

  • Report only what the page returned. If the content does not answer the question, say so rather than filling the gap.
  • Do not fetch pages behind the user's own login, or pages they did not ask for.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.