agentleFS
G4barG4bar/web-cli/skills/web/SKILL.md

web

Skill0 starsChanged 5 days agoMITView on GitHub

Search the web, read pages by URL and get cited answers with the `web` CLI (Exa, automatic TinyFish fallback). Use for current information, docs, library or API lookups, news, papers, and whenever you need the content of a URL. Also has a paid `research` command, only for when the user explicitly asks for deep research.

13 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 12 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write the query as a natural-language description of the page you want, not bare keywords (Exa is semantic).

    line 49

“Common” is measured: kept by at least 6 in 10 of the 12 edited versions of this file we could read. “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

web

CLI on PATH (web). Run it through bash. Output is compact text on stdout; provider errors and fallbacks go to stderr.

Commands

web search "query" [-n 8] [--site a.com,b.org] [--exclude x.com] [--recent 7d] [--cat news|papers|company|people|personal|financial] [--goal "what for"]
web fetch  <url...> [--max 8000] [--focus "what you need"] [--fresh] [--links]
web answer "question" [--deep [--fast]] [--schema file.json|'{...}']
  • search: ranked results, each with title, URL, date and a short excerpt. Start here, then fetch only the 1-3 pages worth reading.
  • fetch: page as markdown, up to 10 URLs. Truncated at --max chars (the header says when). Prefer --focus "..." over raising --max: it keeps only the matching passages.
  • answer: short cited answer for a factual question. --deep = slower, multi-step synthesis (--fast for lighter); --schema = structured JSON answer. Falls back to plain search results when only TinyFish is available.

Options worth knowing

FlagUse
--recent 12h|7d|2w, --since/--until YYYY-MM-DDfreshness (for --cat papers on TinyFish: by publication year)
--site / --excluderestrict or drop domains (comma-separated). Don't put site: in the query
--type lite|auto|deep-lite|deepExa only. lite = fast and cheap, deep* = slower multi-step search; still returns results only, no synthesis. Default auto
--goal "..."say what you are after (search, fetch): helps ranking / extraction
--freshfetch: live page, not cached. --links: also list the page's outbound links
--country US, --lang enlocalise results (--lang is TinyFish only)
-p exa|tinyfishforce one provider, no fallback
--jsonnormalized JSON, for piping into jq

Research (PAID: only when the user explicitly asks for deep research)

web research "question" --max-cost 1 [--effort low|medium|high] [--detach]
web research status <id>    web research cancel <id>
  • Costs real money and takes minutes. --max-cost <usd> is mandatory: use the amount the user gave, never invent one (Exa needs >= 1; TinyFish fallback costs $2-3).
  • Blocks up to --wait minutes (default 15), progress on stderr. Exit 4 = still running: tell the user the printed id, resume with web research status <id>. --detach returns the id immediately.
  • For a quick factual answer use web answer instead (cents, seconds).

Tips

  • Write the query as a natural-language description of the page you want, not bare keywords (Exa is semantic).
  • Save big output instead of printing it: web fetch URL --max 100000 > /tmp/page.md, then rg/read what you need.
  • Exit codes: 0 ok, 1 all providers failed, 2 bad usage or request rejected by the provider (no fallback, fix the flags), 3 fetch partially failed (stderr lists the URLs), 4 research still running.
  • Fallback order is Exa then TinyFish (WEB_PROVIDERS env overrides). A provider whose key is missing is skipped. Warnings on stderr say which flags TinyFish could not honour.
  • web --help shows every option.

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.