web-fetch

name	web_fetch
description	Fetch a URL and return its page content as clean Markdown with a token budget, so an agent can read a web page without burning context. Use the web_fetch tool when you have a specific URL and need its content (not search results). Skill file at backend/.agentium/skills/web_fetch/SKILL.md.
skill_type	research
domain	general
complexity	beginner
tags	["web","fetch","scraping","markdown","retrieval"]
creator_tier	head

Web Fetch

Retrieve the contents of a specific URL as Markdown.

Steps

1. Call the web_fetch tool with action="fetch" and the target url.
2. Set max_tokens to keep the returned Markdown within your context budget.
3. Enable use_cache (default) to avoid re-fetching within 5 minutes.
4. Pass allowed_domains to restrict fetches to trusted hosts.

Validation

• The tool returns status: success with markdown, title, and token_count.
• Oversized pages are truncated and flagged with truncated: true.
• Failures return status: error without raising into the agent context.