Skip to main content
Glama
@mate-tools/mcp-server

@mate-tools/mcp-server

Official
by matetools

@mate-tools/mcp-server

npm version npm downloads Node.js License: MIT

Model Context Protocol server for mate.tools. 35 utility tools for AI agents — text analysis, encoding, hashing, password generation, JSON/CSV/XML/sitemap parsing, regex, color, date, finance, URL metadata extraction, SEO meta-tag generation, text cleanup, CSPRNG random generators, hashtag tools, DNS lookup, SSL/TLS certificate inspection, JWT decoding and more. Wraps the public mate.tools JSON API so Claude Desktop, Claude Code, Cursor, Windsurf, and any other MCP-compatible client can call them directly.

Zero dependencies · Free · No API key · CORS open · Node 18+

Quick install

Add this to your Claude Desktop / Claude Code / Cursor / Windsurf config:

{
  "mcpServers": {
    "mate-tools": {
      "command": "npx",
      "args": ["-y", "@mate-tools/mcp-server"]
    }
  }
}

Restart your client and all 35 tools appear in the picker. That's it.

Related MCP server: CryptoToolBox MCP Server

All 35 tools, by category

Text (8)

Tool

What it does

count_lines

16 text metrics: lines, words, sentences, paragraphs, characters, duplicates, line endings (LF/CRLF/CR), longest word, reading & speaking time.

case_convert

Convert text into 9 case styles in one call: upper, lower, title, sentence, snake, kebab, camel, pascal, constant.

slugify

Convert any text to a URL-safe slug. Unicode-aware, configurable separator/length, optional transliteration of non-Latin characters.

sort_lines

Sort lines: asc/desc, case-sensitivity, dedupe, natural-sort (line2 before line10), locale-aware collation.

lorem_ipsum

Generate placeholder text — paragraphs, sentences or words.

regex_test

Match / replace / split text with a PCRE regex. Returns capture-group offsets. ReDoS-protected.

text_diff

Diff two strings by line, word or character. Returns structured changes + unified diff string + Jaccard similarity.

text_cleanup

Chained cleanup pipeline: trim, collapse_spaces, dedupe_lines, strip HTML/emoji/URLs/emails, normalize quotes/dashes, lower/uppercase.

Encoding (4)

Tool

What it does

base64_encode

Base64-encode text. Standard or URL-safe alphabet (RFC 4648 §5).

base64_decode

Base64-decode. Auto-fixes missing padding; returns hex when not valid UTF-8.

url_encode

Percent-encode a string. Component / form / path mode.

url_decode

Decode percent-encoded URL strings.

Crypto & auth (5)

Tool

What it does

hash

MD5, SHA-1, SHA-256, SHA-384, SHA-512, SHA3-256, SHA3-512, CRC32 (or any subset of PHP's hash algorithms).

password

Generate cryptographically random passwords (CSPRNG). Configurable length, count, character classes, ambiguous-char exclusion. Reports entropy.

password_strength

Score a password 0–4 (zxcvbn-style), entropy bits, offline-GPU crack-time estimate, concerns and suggestions.

credit_card_validate

Luhn check + brand detection (Visa, MC, Amex, Discover, JCB, Diners, UnionPay, Maestro, RuPay). PAN never logged.

jwt_decode

Decode a JWT — header, payload, expiry status. Verify HS256/384/512 with a secret or RS256/384/512 with a PEM public key. Token never logged.

Data & structure (13)

Tool

What it does

json_format

Validate, pretty-print, minify or analyse a JSON document. Returns structural summary (node counts, max depth, top keys).

json_to_csv

Convert a JSON array of objects to CSV. Configurable delimiter, columns, optional flattening of nested objects (meta.city, meta.zip).

xml_validate

Validate XML well-formedness + optional XSD schema. Errors with line/column. XXE-safe.

sitemap_extract

Fetch + parse XML sitemaps. SSRF-protected (refuses private/loopback/link-local/cloud-metadata IPs). Optional recursive expansion of sub-sitemaps.

stats

Descriptive statistics for a list of numbers: mean / median / mode / stddev / variance / percentiles / IQR / geometric / harmonic.

number_base

Convert integers between bases 2..36 + Roman numerals. Big-int safe via GMP.

finance_calc

Multi-mode financial calculations: loan, compound interest, simple interest, discount, tip, sales tax, ROI, percentage, markup.

aspect_ratio

Compute reduced aspect ratio (16:9, 4:3, 21:9, ...) from width × height, or scale a dimension to a target ratio.

color_convert

Convert any color into hex / rgb / rgba / hsl / hsv / cmyk + closest named color + WCAG luminance and contrast (AA / AAA flags).

url_metadata

Fetch a URL and extract title / description / OpenGraph / Twitter Card / canonical / language / hreflang / JSON-LD / favicon / response headers / element counts. SSRF-protected, 2 MB cap.

meta_tags_generate

Generate a full HTML head block (meta + OpenGraph + Twitter Card + JSON-LD) from title/description/url/image. Returns ready-to-paste HTML + SEO score 0-100 with concrete warnings.

dns_lookup

Resolve DNS records (A, AAAA, MX, TXT, NS, CNAME, SOA, CAA, PTR, SRV) via the system resolver. Includes reverse PTR for A records.

ssl_check

Inspect a domain's TLS certificate: subject, issuer, validity, fingerprint, alt names, full chain, days remaining, negotiated protocol/cipher. SSRF-protected.

Date & time (3)

Tool

What it does

timestamp

Bidirectional epoch ↔ ISO 8601 ↔ RFC 3339 ↔ human. Auto-detects epoch resolution. Timezone-aware. Accepts natural language ("next monday", "+2 weeks").

date_math

Add/subtract a duration from a date, or compute the diff between two dates. Optional business-day count (Mon–Fri).

age_calc

Calculate age in years/months/days from DOB. Returns next birthday, days until, Western zodiac sign, generation label.

Generators & social (2)

Tool

What it does

random_gen

CSPRNG-backed generator: number, string (alpha/alphanumeric/numeric/hex/base58/base64url), word, picker (with/without replacement), team balancer, dice (NdM), coin, country, UUID v4 / v7.

hashtag_tools

Hashtag operations on text: extract (with offsets), remove, generate (stop-word-filtered word frequency), format (camel/snake/lower/preserve). Unicode-aware.

The catalog grows over time. npx -y always fetches the latest published version — your client just needs a restart to pick up new tools.

Install snippets for every major client

Claude Desktop

Open the Claude Desktop config:

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

  • Windows: %APPDATA%\Claude\claude_desktop_config.json

  • Linux: ~/.config/Claude/claude_desktop_config.json

Add the mcpServers entry from the Quick install above and restart.

Claude Code

claude mcp add mate-tools npx -- -y @mate-tools/mcp-server

Cursor

Add to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "mate-tools": {
      "command": "npx",
      "args": ["-y", "@mate-tools/mcp-server"]
    }
  }
}

Windsurf and other MCP clients

The server speaks MCP over stdio with the standard JSON-RPC 2.0 framing. Any MCP-compatible client can launch it as a subprocess with the same configuration shape.

Manual install (no npx)

npm install -g @mate-tools/mcp-server
mate-tools-mcp   # speaks JSON-RPC over stdio; plug into a client

How to verify it's working

Once installed, ask your AI agent any of these to confirm the tools are wired:

"Use mate-tools to count the lines and reading time of this paragraph."

"Hash the string secret with SHA-256."

"Generate 10 strong passwords, 20 chars, no ambiguous characters."

"Score this password: correct horse battery staple."

"Convert #ff5733 to all color spaces and tell me the WCAG contrast vs white."

"What's the aspect ratio of 3840 × 2160?"

"Diff these two paragraphs for me."

"What's the monthly payment on a $300k loan at 6.5% APR over 30 years?"

"How old is someone born on 1990-04-15? What's their zodiac sign?"

"Extract all URLs from https://example.com/sitemap.xml recursively."

"Fetch the metadata and OpenGraph tags from https://example.com for me."

"Generate SEO meta tags for an article titled … and score the result."

"Roll 4d6 drop lowest, six times, for me."

"Suggest hashtags for this caption."

"Look up the MX and TXT records for example.com."

"Is the SSL certificate on github.com valid? When does it expire?"

"Decode this JWT and tell me if it's expired."

The agent will pick the right tool automatically.

How it works

The server is a single ~600-line zero-dependency Node script. Every tool call:

  1. Receives a JSON-RPC tools/call request from the MCP client.

  2. Forwards the arguments to the public mate.tools API at https://mate.tools/api/v1/{tool}.php.

  3. Returns a human-readable summary plus the full JSON response back to the client.

Nothing runs locally besides the JSON-RPC plumbing. Computation runs on mate.tools' servers behind Cloudflare. Every endpoint is rate-limited at 60 req/min and 600 req/hour per IP.

Configuration

Override the API endpoint (e.g. for testing against a staging deployment):

MATE_TOOLS_API_BASE=https://staging.mate.tools/api/v1 npx -y @mate-tools/mcp-server

Privacy

Each tool call:

  • ✅ Sends only the arguments you provide to the MCP tool.

  • ✅ Hits HTTPS in transit.

  • ✅ Is processed in memory and not logged with content.

  • ❌ Does not collect telemetry.

  • ❌ Does not require an account, API key or sign-in.

The mate.tools server logs only request method, hashed-IP (rotating daily salt), endpoint name and duration — never request content.

credit_card_validate and password_strength are explicitly engineered so the sensitive input never leaves request-handler memory. Only the validation outcome and metrics are returned.

sitemap_extract, url_metadata, and ssl_check refuse to fetch / connect to any host whose name resolves to a private (RFC 1918), loopback, link-local, or cloud-metadata IP. For HTTP-based tools, each redirect destination is re-validated, so SSRF tricks via 302 don't work. url_metadata additionally caps the response body at 2 MB. ssl_check uses a 5-second connect timeout.

jwt_decode never logs the token. Decoding runs entirely in request-handler memory; only the algorithm name, signature verification outcome, and request duration are written to the per-day access log. Use it confidently with production tokens.

regex_test enforces a hard PCRE backtrack ceiling (100,000 steps) per request — catastrophic regex patterns return a clean 422 rather than hanging the server.

random_gen is backed by PHP's random_int / random_bytes (CSPRNG) — never the Mersenne Twister. UUIDs are RFC 4122 v4 with proper version/variant bits, or v7 with millisecond-precision timestamp.

Source and documentation

Issues / feedback

Open an issue at the mate.tools contact form — please mention "@mate-tools/mcp-server" in the subject line. New tools land roughly every 1–2 weeks; if there's a specific utility you want exposed, just ask.

License

MIT — free for personal and commercial use. See LICENSE.

Available Tools

35 tools
age_calcA

Compute age in years/months/days from a date of birth. Returns next birthday, days until, Western zodiac sign, generation (Gen Z / Millennial / Gen X / Boomer / Silent / Greatest / Gen Alpha), birth weekday, and leap-year info.

Inputs: date_of_birth (required), timezone, reference.

aspect_ratioA

Compute the reduced aspect ratio (16:9, 4:3, 21:9, ...) from a width and height; OR scale a given width OR height to a target ratio.

Inputs: ratio, width, height.

base64_decodeA

Base64-decode a string. Accepts both standard and URL-safe Base64; missing padding is auto-fixed. Returns text when valid UTF-8, otherwise hex.

Inputs: encoded (required).

base64_encodeA

Base64-encode a string. Optional URL-safe variant (RFC 4648 §5: -/_ instead of +//, no padding).

Inputs: text (required), url_safe.

case_convertA

Convert text into multiple case styles in one call: upper, lower, title, sentence, snake, kebab, camel, pascal, constant. Pass cases to limit the response.

Inputs: text (required), cases.

color_convertA

Convert any color (hex, rgb, rgba, hsl, hsla, named) into all standard color spaces. Returns hex, rgb, rgba, hsl, hsv, cmyk, closest CSS named color, WCAG luminance and contrast ratios vs. white and black (with AA/AAA flags).

Inputs: color (required).

count_linesA

Count lines, words, sentences, paragraphs and characters in a chunk of text. Returns sixteen text metrics including duplicate-line detection, longest word, reading and speaking time estimates, and the dominant line-ending style (LF, CRLF or CR).

Inputs: text (required).

credit_card_validateA

Validate a card number using the Luhn algorithm and detect the brand (Visa, MasterCard, Amex, Discover, JCB, Diners, UnionPay, Maestro, RuPay). The card number is never logged — only its length, IIN prefix, last 4, masked form, and Luhn outcome leave the server.

Inputs: number (required).

date_mathA

Date arithmetic. Two modes: provide start + add (e.g. "+90 days", "-2 weeks") to compute the resulting date, OR provide start + end to compute the diff. Diff mode optionally counts business days only.

Inputs: start (required), add, end, business_days.

dns_lookupA

Resolve DNS records for a domain via the system resolver. Default lookup covers A, AAAA, MX, TXT, NS, CNAME, SOA, CAA. Also resolves reverse PTR records for A results. Use this to debug mail config (MX), domain ownership signals (TXT), name-server delegation (NS), CAA policy, and IP resolution. No SSRF risk — DNS only, no outbound connections.

Inputs: domain (required), types.

finance_calcB

Multi-mode financial calculations. mode selects: loan (principal+rate+term [+extra]), compound_interest (principal+rate+years [+contribution]), simple_interest, discount, tip, sales_tax, roi (gain+cost), percentage (sub_mode: of / is_what_percent / change / increase), markup (cost + percent).

Inputs: mode (required).

hashB

Compute one or more cryptographic hashes of a text input. Defaults to MD5, SHA-1, SHA-256, SHA-384, SHA-512, CRC32, SHA3-256, SHA3-512.

Inputs: text (required), algorithms.

hashtag_toolsA

Hashtag operations on text. mode=extract (pull #hashtags out of text with offsets), remove (strip them out + tidy whitespace), generate (suggest #hashtags from text content via stop-word-filtered word frequency), format (normalize an array of hashtag strings to camel/snake/lower/preserve style). Unicode-aware (#日本語 works).

Inputs: mode (required), max, text, style, +2 more.

json_formatA

Validate, pretty-print, minify or analyse a JSON document. Returns a structural summary (node counts, max depth, top distinct keys).

Inputs: json (required), mode, indent.

json_to_csvA

Convert a JSON array of objects to CSV. Configurable delimiter, columns, optional flattening of nested objects with dot-keys (e.g. meta.city). Max 50,000 rows per request.

Inputs: json (required), columns, flatten, newline, +2 more.

jwt_decodeA

Decode a JSON Web Token (JWT) and analyse its claims. Always returns header, payload, signature info, and expiry status (active/expired/not_yet_valid) computed from exp/nbf/iat. When secret is provided, verifies HS256/HS384/HS512 via HMAC. When public_key is provided in PEM format, verifies RS256/RS384/RS512 via openssl_verify. ES* and EdDSA are decoded but not verified. The token is never logged.

Inputs: token (required), secret, public_key.

lorem_ipsumA

Generate placeholder text — N paragraphs, sentences or words of lorem ipsum. Suitable for mocks, fixtures and test data.

Inputs: kind, count, start_with_lorem.

meta_tags_generateA

Generate a complete HTML head block — standard , OpenGraph, Twitter Card and Schema.org JSON-LD — from a small set of inputs (title, description, url, image, author, keywords, type). Returns ready-to-paste HTML plus an SEO score 0-100 with concrete warnings (title length, description length, missing OG image, etc.).

Inputs: title (required), description (required), url, type, +7 more.

number_baseA

Convert integers between any base 2..36 plus Roman numerals. Big-int safe when GMP is available. Use from_base="roman" to parse a Roman numeral input, or include "roman" in to_bases to convert to Roman.

Inputs: value (required), to_bases, from_base.

passwordA

Generate cryptographically random passwords using a CSPRNG. Configurable length, count, character classes and ambiguous-character exclusion.

Inputs: count, digits, length, symbols, +3 more.

password_strengthA

Score a password (0-4 zxcvbn-style), compute entropy bits, identify common-password fingerprints and keyboard sequences, and estimate offline-GPU crack time. Password is never logged.

Inputs: password (required).

random_genA

Cryptographically random generator backed by random_int / random_bytes. mode=number (min/max), string (length + charset: alpha/alphanumeric/numeric/lower/upper/hex/hex_upper/base58/base64url), word (English word list), picker (pick N from a list, with/without replacement), team (balanced round-robin team allocation), dice (NdM notation), coin, country (ISO list), uuid (v4 or v7).

Inputs: mode (required).

regex_testA

Match, replace or split text with a PCRE regular expression. Returns offsets and capture groups. ReDoS-protected: catastrophic patterns return an error rather than hanging.

Inputs: text (required), pattern (required), flags, operation, +2 more.

sitemap_extractA

Fetch and parse an XML sitemap or sitemap index from a public HTTPS URL (or paste raw XML). SSRF-protected: refuses private/loopback/link-local/cloud-metadata addresses, re-validates redirects. Optional recursive expansion of sub-sitemaps. Returns lastmod, changefreq, priority when present.

Inputs: url, xml, max_urls, recursive, +2 more.

slugifyA

Convert any text to a URL-safe slug. Unicode-aware: by default transliterates non-Latin characters to ASCII (set transliterate=false to keep them). Configurable separator and length.

Inputs: text (required), lowercase, separator, max_length, +1 more.

sort_linesA

Sort lines of text. Supports asc/desc, case-insensitive, locale-aware collation, natural-sort (line2 before line10), dedupe, line-trim and empty-line drop.

Inputs: text (required), order, dedupe, locale, +4 more.

ssl_checkA

Connect to {domain}:{port} over TLS and return the leaf certificate + full chain: subject, issuer, validity dates, days remaining, fingerprint (SHA-256), subject alternative names, signature algorithm, negotiated protocol and cipher. Useful for verifying a domain has HTTPS, the cert is not expired, and the chain is properly configured. SSRF-protected: refuses private/loopback/link-local/cloud-metadata IPs. 5s connect timeout.

Inputs: domain (required), port.

statsA

Descriptive statistics for a list of numbers: count, sum, mean, median, mode, min/max/range, variance, stddev (population AND sample), MAD, quartiles, IQR, percentiles (1-99), geometric mean, harmonic mean.

Inputs: numbers (required).

text_cleanupA

Apply chained text cleanup operations in sequence. Operations: trim, collapse_spaces, normalize_whitespace, remove_line_breaks, normalize_line_breaks, dedupe_lines, reverse_lines, strip_html, strip_emoji, strip_urls, strip_emails, normalize_quotes (curly → straight), normalize_dashes (em/en → hyphen), lowercase, uppercase. Returns cleaned text plus per-operation chars/lines removed.

Inputs: text (required), operations (required).

text_diffA

Diff two strings by line, word or character. Returns structured changes (+/-/=), git-style unified diff, and a Jaccard similarity score. Use line mode for code/config; word mode for prose; char mode for short strings.

Inputs: a (required), b (required), mode, context.

timestampA

Convert any moment in time between formats. Bidirectional Unix epoch ↔ ISO 8601 ↔ RFC 3339 ↔ human-readable. Auto-detects epoch resolution. Accepts strtotime-style natural language ("next monday", "+2 weeks"). Optional timezone for local components.

Inputs: input, timezone.

url_decodeA

Decode a percent-encoded URL string. mode=component (rawurldecode, default) or form (interprets + as space).

Inputs: encoded (required), mode.

url_encodeA

Percent-encode a string for URL use. mode=component (RFC 3986 strict, default), form (application/x-www-form-urlencoded), or path (keeps slashes intact).

Inputs: text (required), mode.

url_metadataA

Fetch a public HTTPS URL and extract structured page metadata: title, description, keywords, canonical, language, charset, full OpenGraph dictionary, Twitter Card, favicon (absolute URL), hreflang map, JSON-LD blocks, response headers and counts (h1/h2/images/links/scripts). SSRF-protected: refuses private/loopback/link-local/cloud-metadata addresses and re-validates every redirect. 2 MB body cap.

Inputs: url (required), timeout, user_agent, follow_redirects.

xml_validateA

Validate XML well-formedness; with optional XSD schema. Reports errors with line/column. Network and DTD fetches are disabled (no XXE risk).

Inputs: xml (required), schema.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.6/5.0

    Scored across 35 tools

    Disambiguation4/5

    Each tool targets a distinct operation, but `text_cleanup` and `case_convert` both handle case changes (lowercase/uppercase), and `sitemap_extract` vs `url_metadata` both involve fetching URLs, so a careful reading is needed to pick the right tool. Overall, no severe overlaps.

    Naming Consistency3/5

    Tool names are all lowercase with underscores, but the pattern is inconsistent: many are noun_verb (`base64_encode`, `url_decode`, `regex_test`), some are single nouns (`hash`, `password`, `stats`), some use abbreviations (`age_calc`, `finance_calc`, `random_gen`), and a few don't follow verb_noun (`lorem_ipsum`, `json_to_csv`, `hashtag_tools`). This mixed style is still readable but not uniform.

    Tool Count2/5

    With 35 tools, the server is overloaded for a single MCP server. Although the scope is a general utility toolkit, the number exceeds what an agent can efficiently navigate, and many tools could be split into focused servers. This borders on being too many and earns a score of 2.

    Completeness4/5

    The toolkit covers a wide range of common utilities: encoding, text processing, date/time, math, finance, web metadata, validation, and security. Minor gaps exist (e.g., no image processing, no file system operations), but for a general-purpose utility server, the surface is quite complete and leaves no major dead ends for typical use cases.

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      A
      quality
      C
      maintenance
      Swiss-army-knife utility MCP server for AI agents. 18 tools for JSON validation/formatting, base64 encode/decode, hash generation, UUID generation, URL parsing, regex testing, markdown↔HTML conversion, text stats, slug generation, datetime conversion, cron parsing, text diffing, CSV↔JSON conversion, and JWT decoding. Zero API Key required
      18
      5
      -
    • A
      license
      Not graded
      quality
      D
      maintenance
      An MCP server that exposes 83 utility tools (string, JSON, CSV, list, time, number, XML) to AI agents, auto-generated from the omni-tools source.
      MIT
    • A
      license
      A
      quality
      D
      maintenance
      An MCP server that bundles 53 online utility tools (164 capabilities) for AI assistants, covering JSON processing, encryption/decryption, code formatting, text conversion, unit conversion, network queries, QR/barcode generation, and reference lookups via natural language calls.
      53
      1
      MIT