# Agentmarginrouter Agent Margin Router MCP server

Source: https://glimind.com/server/mcp-registry%2Fio.github.agentmarginrouter%2Fagent-margin-router
Registry id: `mcp-registry/io.github.agentmarginrouter/agent-margin-router`
Status: live (checked 2026-10-27T06:02:17.803Z)

## Verdict

Agentmarginrouter Agent Margin Router MCP server answered 99.97% of Glimind's liveness checks over 30 days, better than 40% of MCP servers, with 629 ms median handshake and 1 outage in 90 days. Its agent-readiness grade is A (96/100) across 53 tools, costing an agent about 13,864 tokens to load. Checked every few minutes; no tool is ever called.

## Measurements (30 days)

| Metric | Value | Better than |
|---|---|---|
| Uptime, 30 days | 99.97% | 40% of MCP servers |
| Median handshake latency | 629 ms | 36% of MCP servers |
| p95 handshake latency | 2610 ms | — |
| Checks in 30 days | 4,991 | — |
| Outages, 90 days | 1 | — |
| Median recovery | 6 min | — |
| Agent-readiness score | 96/100 (A) | 82% of MCP servers |
| Tools | 53 | — |
| Tool-list size for an agent | ~13,864 tokens | 13% of MCP servers |
| Days observed | 29 | — |

## Install

```json
{
  "mcpServers": {
    "agentmarginrouter-agent-margin-router-mc": {
      "type": "http",
      "url": "https://agent-margin-router-production.up.railway.app/mcp"
    }
  }
}
```

Claude Code:

```sh
claude mcp add --transport http agentmarginrouter-agent-margin-router-mc https://agent-margin-router-production.up.railway.app/mcp
```

## Tools (53)

- `agent_wallet_snapshot`: One-call wallet overview for autonomous agents: native balance, ERC-20 balances (default stablecoins + WETH, or any contracts you pass), current gas price, tx c
- `arxiv_search`: Search arXiv preprints: returns id, title, authors, abstract, categories, published/updated dates and PDF links, sortable by relevance or date, optionally filte
- `best_price`: Best-price router: queries several independent gas-price sources in parallel, returns the cheapest quote and charges a configurable share of the savings (averag
- `contract_info`: Smart-contract metadata from Blockscout for Base, Ethereum, Optimism or Arbitrum: verification status, name, compiler, proxy/implementation, creator, creation t
- `crypto_ohlcv`: Historical OHLCV candles for any major crypto pair (Binance, Coinbase fallback): intervals 1m-1w, up to 500 candles, with period summary (high, low, volume, cha
- `currency_convert`: Fiat currency conversion with ECB reference rates (30+ currencies): amount, source currency, one or many targets. Returns rates, converted amounts and rate date
- `defi_tvl`: Total value locked of a DeFi protocol (by DefiLlama slug: TVL, 1d/7d change, market cap, mcap/TVL ratio, TVL per chain, category) or of a whole chain (TVL, nati
- `defi_yields`: Live DeFi pool yields (APY, TVL, stablecoin flag) across 100+ chains and protocols from DefiLlama. Filter by chain, asset, project, minimum TVL; sorted by APY. 
- `dex_price`: Live DEX price, 24h change, volume and liquidity for any on-chain token (long-tail tokens not listed on CEXs) by contract address or symbol on Base, Ethereum, A
- `dns_lookup`: Resolve any DNS record type (A, AAAA, CNAME, MX, NS, TXT, SOA, SRV, CAA, PTR) over DNS-over-HTTPS with TTLs and DNSSEC validation flag. Pay per call with USDC o
- `dns_whois`: DNS records (A, AAAA, MX, NS, TXT, CNAME, SOA, CAA) via DNS-over-HTTPS plus RDAP/WHOIS registration data (registrar, created/expires, status, nameservers, DNSSE
- `domain_rdap`: Domain registration data via RDAP (the WHOIS successor): registered flag, registrar, status, registration/expiry events, nameservers and DNSSEC – for availabili
- `email_validate`: Validate an email address: RFC syntax, MX/A records via DNS-over-HTTPS, disposable and free-provider detection, role-account flag, deliverability score and verd
- `ens_resolve`: Resolve an ENS name to its Ethereum address plus text records (url, com.twitter, avatar, ...) via on-chain calls. Pay per call with USDC or USDT on Base via x40
- `erc20_balance`: Balances of up to 20 arbitrary ERC-20 tokens for one wallet in a single Multicall3 round-trip: raw and decimal-formatted balance, decimals and symbol per token,
- `extract_clean`: Clean page extraction from any public URL: title, meta description, main text, JSON-LD, language. Optional `schema` keeps only the listed top-level keys. Apify 
- `fear_greed`: Crypto market sentiment: the daily Fear & Greed index (0-100) with classification, 1d/7d change and up to 90 days of history from alternative.me. Pay per call w
- `feed_discover`: Find the RSS/Atom feed URLs for a website from its HTML link tags, with a common-path fallback (/feed, /rss.xml, ...). Pay per call with USDC or USDT on Base vi
- `gas_fees`: Live EIP-1559 gas fees (slow/standard/fast tiers) plus USD cost per transaction type (transfer, ERC-20 transfer, swap, NFT mint) for Base, Ethereum, Arbitrum, O
- `geo_ip`: IP geolocation for IPv4/IPv6: country, region, city, coordinates, timezone, ASN, ISP/org, EU flag, private-range detection. Pay per call with USDC or USDT on Ba
- `geocode`: Turn a place name or address into coordinates, bounding box, address parts and OSM type using OpenStreetMap Nominatim (ODbL); cached 24h. Pay per call with USDC
- `github_repo_summary`: One-call overview of any public GitHub repository: description, stars, forks, open issues, primary language and language split (%), topics, license, default bra
- `hash`: Compute cryptographic hash digests (MD5, SHA-1, SHA-224, SHA-256, SHA-384, SHA-512) of any text. Key-less, local computation. Pay per call with USDC or USDT on 
- `health`: Free. Returns AMR service status, provider health and cache metrics.
- `hn_search`: Search Hacker News stories and comments via the Algolia HN API: title, URL, author, points, comment count, timestamp and HN discussion link; sort by relevance o
- `holidays`: Official public holidays for a country and year via Nager.Date (key-less): date, local + English name, whether it is nationwide, and sub-region counties. Pay pe
- `html_to_markdown`: Convert any public web page into clean, LLM-ready Markdown: main content only (boilerplate, nav and ads removed), headings, lists, tables and optional links/ima
- `market_spread`: Cross-exchange crypto spread from live tickers: gross spread, fees, modelled slippage, transfer cost and net edge in bps for a given trade size. Ticker/volume b
- `media_convert`: Convert or transcode a public audio/video/image file via a self-hosted FFmpeg container: pass a source URL and a target format (mp3, mp4, webm, wav, png, …) and
- `metadata`: Structured metadata for any public page in one call: title, meta description/keywords, canonical URL, HTML lang, charset, favicon (absolute URL) and the FULL Op
- `news_search`: Global news article search across 65+ languages via the GDELT DOC 2.0 API: title, URL, source domain, publication time, language and country per article, with l
- `npm_package`: npm package intelligence: latest version, license, repository, maintainers, dependency count, publish dates and weekly downloads – ideal for dependency vetting 
- `payment_info`: Free. Returns the x402 payment requirements (price, USDC/USDT contract, pay-to wallet, network) for a given tool so an agent can construct an x402 payment.
- `pdf_text`: Extract clean text (per page + full) and metadata from any public PDF URL up to 25 MB / 200 pages. Pay per call with USDC or USDT on Base via x402. 3 free calls
- `pypi_package`: PyPI project intelligence: latest version, summary, license, author, project URLs, Python requirement, declared dependencies, release count and last release dat
- `qr_generate`: Generate a QR code from any text or URL and return it as a scalable SVG image plus a data-URI. Pure key-less generation, no third-party service. Pay per call wi
- `reddit_search`: Search Reddit posts site-wide or within one subreddit: title, URL, subreddit, author, score, upvote ratio, comment count, timestamp and the first 500 characters
- `robots_check`: Answer 'may this user-agent crawl this URL?' by fetching and parsing the site's robots.txt: allowed flag, crawl-delay, request-rate and declared sitemaps. Pay p
- `rss_feed`: Fetch and parse any RSS 2.0 or Atom feed into clean JSON: title, link, published date, author, summary, optional full text content and categories per item. Pay 
- `seo_check`: On-page SEO audit of any URL: title, meta description, canonical, headings, images without alt, links, Open Graph, structured data, robots.txt/sitemap, prioriti
- `sitemap_urls`: Parse a sitemap into the real URL list: accepts a direct sitemap.xml, a <sitemapindex> (recursed, bounded) or a site root (robots.txt Sitemap: lines, then /site
- `text_stats`: Readability and text statistics for raw text or a fetched page: word/sentence/syllable counts, Flesch Reading Ease, Flesch-Kincaid grade, and reading/speaking t
- `timezone`: Current local time, UTC offset, DST status and abbreviation for any IANA timezone name, or resolve a place name to its timezone and local time. Pay per call wit
- `token_price`: Multi-venue median spot price for any major crypto asset (Binance, Coinbase, Kraken, CoinGecko) with min/max, deviation and confidence score – one call instead 
- `tx_status`: Status of any EVM transaction on Base, Ethereum, BSC, Polygon, Arbitrum or Optimism: success/failed/pending, block number, confirmations, gas used, effective ga
- `url_status`: Health check for any public URL: HTTP status, full redirect chain, response time, content type, SSL certificate validity/expiry/issuer and a security-header aud
- `uuid`: Generate one or more RFC 4122 UUIDs (random v4, or deterministic v5 from a namespace and name). Key-less, local generation. Pay per call with USDC or USDT on Ba
- `wallet_balance`: Native + stablecoin balances (ETH, USDC, USDT, DAI, WETH) of any EVM wallet with USD valuation, read directly from public RPC nodes on Base, Ethereum, Arbitrum,
- `weather`: Current conditions plus a daily forecast (up to 16 days) for any place name or lat/lon via Open-Meteo (key-less, CC BY 4.0); temperature in °C or °F, cached 30 
- `web_crawl`: Mini crawler: fetches a start page and follows same-domain links breadth-first (max 10 pages), returning clean LLM-ready Markdown, title, word count and discove
- `web_search`: Live web search results (title, URL, domain, snippet) for any query, key-less, region-aware. Pay per call with USDC or USDT on Base via x402. 3 free calls per w
- `wikipedia_summary`: Clean article summary from Wikipedia in any language: title, short description, extract (capped), canonical URL, thumbnail and coordinates; falls back to search
- `youtube_transcript`: Fetch the transcript of a YouTube video (manual or auto-generated captions) as timestamped segments plus full text, with language preference and English fallbac

Measured by Glimind (https://glimind.com): MCP handshake + tool list only, no tool is ever called. CC BY 4.0.