Kael MCP Server

Encode text to Base64 or decode Base64 to text. Use mode='encode' (default) or mode='decode'.

Execute code in a sandboxed environment and return stdout/stderr. Supports JavaScript (Node.js), Python 3, and Bash. 5-second timeout, 50KB output limit. Much cheaper than the LLM simulating execution mentally.

Look up DNS records for a domain. Returns A, AAAA, MX, TXT, NS, CNAME, SOA, or SRV records. Real-time data that the LLM cannot know without querying.

Hash text using MD5, SHA1, SHA256, or SHA512. Use algorithm='all' for all hashes at once.

Extract structured data from any webpage: tables (as arrays), lists, headings hierarchy, images with alt text, meta tags, and custom CSS selectors. Returns clean JSON instead of messy HTML. Saves massive token costs vs having LLM parse raw HTML.

Fetch HTTP response headers from a URL without downloading the body. Shows status code, content-type, server, caching headers, etc.

Get geolocation data for any IP address: country, region, city, coordinates, timezone, ISP, organization, AS number. Real-time enrichment data.

Query/extract data from JSON using dot-notation paths. Supports [0] indexing and [] wildcards. Example: 'data.users[].name'

Extract text content from a PDF at a given URL. Returns page count, metadata, and full extracted text.

Test a regex pattern against text. Returns all matches with indices, capture groups, and match count.

Take a screenshot of any webpage. Returns a PNG image as base64. Useful when the AI needs to see what a page looks like without processing the full DOM. Uses headless Chromium/Playwright.

Compare two text strings and return a unified diff showing additions (+), removals (-), and unchanged lines.

Follow URL redirects and return the final destination URL with the full redirect chain. Useful for bit.ly, t.co, etc.

Fetch a web page and extract its readable content as clean text/markdown. Much cheaper than having the LLM process raw HTML. Handles redirects, strips ads/nav/scripts, extracts main content, title, description, and top links.

Search the web and return structured results with titles, URLs, and snippets. Uses DuckDuckGo. Returns real-time search results the LLM cannot access from training data.

WHOIS lookup for any domain. Returns registrar, creation/expiry dates, name servers, status, and registrant info. Real-time data the LLM cannot access without this tool.