Back to Browse

Agent Utilities MCP Server

by Agent Utilities
Developer ToolsLow Risk8.0Remote
Free

Free agent tools: SEO snapshots, sitemap refresh, Wikipedia collections and saved-page MCP guides.

About

Review static SEO findings before and after a site edit, with reusable reading data for agents. All available full hosted tools require no account, key, wallet or payment during validation.

Start at /seo with the URL and supplied static HTML for a page you may inspect. Audit locally in the browser, export a baseline, and compare it with a capture after your next real edit. The supplied URL is a label; this form does not fetch it or invent an HTTP status or source capture time. Supplied HTML stays in your browser. Raw baselines contain the original HTML and should remain private. Comparison reports preserve dates, coverage and unknowns; missing or failed observations do not prove a finding was resolved.

Hosted SEO accepts up to 20 same-site static snapshots within 128 KiB. It returns grouped findings, stable IDs and paginated evidence for HTTP failures, broken internal links, missing titles, descriptions, index directives, canonical conflicts and JSON-LD syntax. It does not retrieve missing pages, render JavaScript, measure rankings, keyword volume or backlinks, or guarantee whole-site coverage.

Local Node.js 22+ packages add bounded same-HTTPS-origin crawling, sitemap discovery and exact captured-URL refresh with compact comparisons. The crawler respects robots.txt and public DNS restrictions: at most 20 captured pages, five sitemap requests, 200 sitemap candidates, 100 KB per response, 2 MB total and 60 seconds.

Collect 1–20 same-edition Wikipedia articles into a private reusable capture. It retains revisions, licenses, retrieval dates, request mappings and source omissions; failed refreshes can retain explicitly stale text. Reads share 8 MB and 60 seconds. Local lexical search and unique exact heading selection work without fetching again.

Optional saved-page MCP: run the downloaded page-library module with serve collection.json. Four readonly local tools list metadata, search, select captured sections and inspect stored outlines. This is a separate stdio connection to one configured snapshot, not extra hosted tools. It performs no source requests during serving; refresh separately and restart. Results include dates/revisions/stale and omission flags, with a 64 KiB serialized tool-result cap.

Guide, downloads, checksums and standard MCP template: https://agent-utilities.agent-utilities.workers.dev/seo?utm_source=mcp-marketplace Machine-readable guide: https://agent-utilities.agent-utilities.workers.dev/v1/content/docs/seo-page-library?utm_source=mcp-marketplace

Optional named-tokenizer payload counts are not model-cost or dollar-savings claims. Other free tools include supplied-HTML extraction and supplied-data Reddit formatting; live Reddit retrieval remains unconfigured. Hosted shared capacity and per-source/input/output bounds apply. Charging will not resume automatically.

Security Report

8.0
Low Risk8.0Low Risk

Remote MCP endpoint verified (205ms response). Server: agent-utilities-remote. 64 tools available. 2 trust signals: valid MCP protocol, known domain (workers.dev). 1 security issue detected.

64 tools verified · Open access · 2 issues found

Security scores are indicators to help you make informed decisions, not guarantees. Always review permissions before connecting any MCP server.

Remote servers are capped at 8.0 because source code is not available for review. The score reflects endpoint verification only.

Permissions Required

This plugin requests these system permissions. Most are normal for its category.

HTTP Network Access

Connects to external APIs or services over the internet.

How to Connect

Remote Plugin

No local installation needed. Your AI client connects to the remote endpoint directly.

Add this to your MCP configuration to connect:

{
  "mcpServers": {
    "dev-workers-agent-utilities-agent-utilities-agent-utilities": {
      "url": "https://agent-utilities.agent-utilities.workers.dev/mcp/remote"
    }
  }
}

Getting Started

Once installed, try these example prompts and explore these capabilities:

  • 1Connect to https://agent-utilities.agent-utilities.workers.dev/mcp/remote?category=web without Authorization for hosted tools.
  • 2Try agent_utilities_seo_audit with direct arguments: {"siteUrl":"https://example.com/","pages":[{"url":"https://example.com/","status":200,"html":"<title>Example shop</title>"}]}. This is a fictional snapshot example, not customer evidence.
  • 3For a permitted public site: node agent-utilities-seo.mjs crawl https://your-site.example/ --sitemaps --out before.json; later run node agent-utilities-seo.mjs refresh before.json --out after.json. Source coverage and omissions matter.
  • 4For Wikipedia, download wikipedia-collection-example.json and run node agent-utilities-pages.mjs collect wikipedia-collection-example.json --out collection.json. The example collects Earth and Moon; it is not a popularity ranking.
  • 5Run node agent-utilities-pages.mjs select collection.json --url https://en.wikipedia.org/wiki/Moon --section "Names and etymology" for the captured section, or use serve collection.json as the local MCP command. Fill the saved-pages-mcp.json template with absolute local paths in your MCP-capable client. Four local tools are distinct from the hosted endpoint; client apps themselves are untested.
  • 6Download instructions, source bounds and provenance: https://agent-utilities.agent-utilities.workers.dev/seo?utm_source=mcp-marketplace . No live Reddit fetching.

Reviews

No reviews yet

Be the first to review this server!

Version History

v1.12.0Oct 6, 2026Re-scanned

Listing metadata 1.12.0 adds verified sitemap discovery/refresh, multi-article Wikipedia collections and a separate readonly saved-library stdio MCP connection. Four saved-library tools are local, not additions to the hosted endpoint. Free access and documented source/capacity bounds remain; no whole-site, ranking, measured savings or live Reddit claim. Reader package remains 1.1.1.

v1.11.0Oct 6, 2026Re-scanned

Listing metadata 1.11.0 adds the deployed free SEO first call, bounded local crawler and reusable page-library guide. All available full tools remain free without credentials or payment. No rankings, backlinks, rendering, whole-site mirror or measured model-cost claims. Live Reddit remains unconfigured; supplied-data formatting works. Reader package remains 1.1.1.

v1.10.0Oct 6, 2026Re-scanned

All available full hosted tools are free during customer validation: no account, key or wallet. Funding and execution charges paused; existing balances preserved. Shared bounded capacity and source-access limits apply. Live Reddit remains unavailable. Reader package stays 1.1.1.

v1.9.1Oct 6, 2026Re-scanned

Updated hosted MCP metadata and free first-call instructions. Unconfigured live Reddit tools are omitted from discovery; supplied-data Reddit formatting remains available. Invalid free-call inputs return schemas and examples without paid fallback. Reader package remains 1.1.1.