Zactonz MCP Server
Retrieves a direct download link for a Google Drive file via the Zactonz gdrive API, so an assistant can obtain a downloadable URL for a shared Drive file.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Zactonz MCP Server take a screenshot of https://example.com"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Zactonz MCP server
A Model Context Protocol server for the Zactonz APIs. It lets an AI assistant capture a screenshot of a web page, read a page as Markdown, preview a link, look up DNS, SSL and WHOIS records, check a domain's email setup, verify addresses, generate QR codes, barcodes and social card images, convert images and translate text.
It runs on your machine over stdio and needs Node.js 18 or newer.
Install
There is nothing to install ahead of time. The client configurations below run
npx -y @zactonz/mcp, which fetches the package on
first use and keeps it cached. You need Node.js 18 or newer.
Skip to Setup unless you want to run it from a checkout.
From source
Useful for pinning a particular commit, working on the server, or running without a registry. Needs Node.js 18 or newer and git.
git clone https://github.com/zactonz/zactonz-mcp.git
cd zactonz-mcp
npm ci --omit=devThat is the whole install: two runtime dependencies, no build step. Check it starts — it will wait for input, so press Ctrl+C to leave:
node bin/zactonz-mcp.jsThen point your client at the absolute path, in place of the npx form used below:
{
"mcpServers": {
"zactonz": {
"command": "node",
"args": ["/absolute/path/to/zactonz-mcp/bin/zactonz-mcp.js"],
"env": {
"ZACTONZ_API_KEYS": "zk_screen_…,zk_markdown_…"
}
}
}
}Use the real absolute path — ~ and relative paths are not expanded by most clients. On Windows,
write it with forward slashes or escaped backslashes, for example
C:/Users/you/zactonz-mcp/bin/zactonz-mcp.js.
For Claude Code:
claude mcp add zactonz --env ZACTONZ_API_KEYS="zk_screen_…" -- node /absolute/path/to/zactonz-mcp/bin/zactonz-mcp.jsIf you would rather have zactonz-mcp on your PATH, run npm link in the clone, then use
"command": "zactonz-mcp" with no args.
To update later: git pull && npm ci --omit=dev.
Related MCP server: rasterly-mcp
Setup
Create a key for each product you want to use in the API console. There is a free plan.
Add the server to your MCP client, with the keys in
ZACTONZ_API_KEYSseparated by commas.
If you installed from source instead, substitute the "command" and "args" shown
there for the npx form used below.
Clients that read an mcpServers block, such as Claude Desktop, Cursor and Windsurf:
{
"mcpServers": {
"zactonz": {
"command": "npx",
"args": ["-y", "@zactonz/mcp"],
"env": {
"ZACTONZ_API_KEYS": "zk_screen_…,zk_markdown_…,zk_domain_…"
}
}
}
}VS Code (.vscode/mcp.json):
{
"servers": {
"zactonz": {
"command": "npx",
"args": ["-y", "@zactonz/mcp"],
"env": { "ZACTONZ_API_KEYS": "zk_screen_…,zk_markdown_…" }
}
}
}Claude Code:
claude mcp add zactonz --env ZACTONZ_API_KEYS="zk_screen_…,zk_markdown_…" -- npx -y @zactonz/mcpOn Windows, if the client cannot start npx directly, use "command": "cmd" with "args": ["/c", "npx", "-y", "@zactonz/mcp"].
Keys
A Zactonz key works for one product, and the product is part of the key: zk_screen_… is a screenshot key, zk_markdown_… a Markdown key. Put every key you have in ZACTONZ_API_KEYS. The server uses the right one for each call and offers the assistant only the tools those keys can call.
Keys stay in the server process. They are sent to api.zactonz.com in the Authorization header and nowhere else, and they are never included in anything returned to the assistant.
Tools
Tool | Purpose | Key |
| Capture a web page as an image or PDF |
|
| Render HTML to an image or PDF |
|
| Read a web page as Markdown |
|
| Convert HTML to Markdown |
|
| Title, description, image and metadata of a URL |
|
| DNS records and DNSSEC status |
|
| Certificate, chain, expiry and TLS details |
|
| Registrar, dates and nameservers of a domain |
|
| MX, SPF, DKIM, DMARC and related records, scored |
|
| Whether addresses can receive mail |
|
| Generate a QR code |
|
| Decode a QR code |
|
| Generate a barcode |
|
| Generate an Open Graph image |
|
| Convert, resize or crop an image |
|
| Dimensions, format and colours of an image |
|
| Translate between 46 languages |
|
| Direct download link for a Google Drive file |
|
| Plan and remaining quota of a key | any |
docs/tools.md lists the arguments of each tool.
Example requests once it is connected:
"Take a screenshot of example.com on a 390 pixel wide screen and tell me whether the menu fits."
"Read https://example.com/pricing and summarise the plans."
"Does example.com have SPF and DMARC set up correctly?"
"When does the SSL certificate for example.com expire?"
Results
A tool returns JSON text. Two things differ from the raw API:
read_webpageandconvert_html_to_markdownreturn the Markdown as plain text, followed by the remaining fields as JSON. They request at most 20,000 characters unless the assistant asks for more.A tool that produces an image returns the link and, when the file is 750 KB or smaller, the image as well, so the assistant can look at it.
capture_screenshotandrender_htmluse JPEG quality 80 unless told otherwise, to stay under that size. PDFs and larger images are returned as a link only.
A refused call is returned as a tool error carrying the API's message, the status, how long to wait when a limit was reached, and the request id.
Limits and timing
Calls spend units from your plan's quota, the same as direct API calls. See rate limits and quotas.
A tool call is given 55 seconds in total, because MCP clients stop waiting after 60. Within that time the server retries once, and only where a retry cannot repeat work: after a 429 with a short Retry-After, after a gateway error on a read, or ten seconds after the API's request burst limit rejects a call. If the client cancels a call, the server stops.
Configuration
Variable | Purpose | Default |
| API keys, separated by commas | none |
| Largest image returned inline, in bytes. |
|
| API base URL |
|
Without keys the server still starts and lists every tool, and each call answers with setup instructions. Values in ZACTONZ_API_KEYS that are not Zactonz keys are ignored and reported on stderr.
Security and privacy
Untrusted content.
read_webpage,preview_linkandread_qr_codereturn text written by whoever controls the page or the code. That text can contain instructions aimed at the assistant. The server labels it as third-party data, but the label is advice to the model, not a guarantee. Review what an assistant does after reading pages you do not control.Generated files are public links. Screenshots, PDFs, QR codes, barcodes and converted images are stored on
api.zactonz.comat unguessable URLs that anyone holding the link can open, for the period given on each endpoint's reference page. Do not render documents that must stay private.What is sent. Tool arguments go to
api.zactonz.comand nowhere else. The server keeps no logs and stores nothing on disk. The API's own handling of data is covered by the privacy policy.Arguments are validated against each tool's schema before any request is made, and arguments outside the schema are refused.
Images are fetched for inline display only from the API's own host, over HTTPS, without following redirects.
Tools that fetch a URL do so from Zactonz's servers, which refuse private and internal addresses.
Report a vulnerability as described in SECURITY.md.
Troubleshooting
A tool is missing. Only tools with a key are listed. Add a key for that product and restart the client.
Every call answers with setup instructions.
ZACTONZ_API_KEYSis empty or holds no valid key. The server prints the reason on stderr, which most clients show in their MCP log."Missing or invalid API key." The key was revoked or mistyped. Ask the assistant to run
check_api_key, or check the key in the console.A screenshot has no inline image. The file is over the inline limit. Lower the quality or size, or raise
ZACTONZ_MCP_INLINE_IMAGE_BYTES.
Versioning
This package follows semantic versioning. While it is at 0.x, a minor release may rename a tool or an argument; such changes are listed in the changelog.
Development
See CONTRIBUTING.md.
Licence
MIT. See LICENSE. Maintained by Zactonz Technologies.
Available Tools
19 toolscapture_screenshotCapture a screenshot of a web pageA
Loads a public web page in a headless browser and captures it as a JPEG, PNG or WebP image, or as a PDF. Returns a link to the file, and the image itself when it is small enough to look at directly. Suited to checking how a page renders or keeping a visual record of it.
Inputs: url (required), delay, scale, width, +10 more.
check_api_keyCheck a Zactonz API keyARead-only
Reports the plan of a configured Zactonz API key and the quota it has left today and this month. Costs nothing.
Inputs: product (required).
check_email_domainCheck the email setup of a domainBRead-only
Audits how a domain is configured for email: MX, SPF, DKIM, DMARC, MTA-STS, TLS-RPT and BIMI records, whether it is a disposable or free mail provider, a score out of 100 and a list of findings.
Inputs: domain (required), probe, selectors.
convert_html_to_markdownConvert HTML to MarkdownBRead-only
Converts HTML you already hold to Markdown. Nothing is fetched.
Inputs: html (required), mode, links, images, +1 more.
convert_imageConvert or resize an imageA
Converts an image at a public URL to WebP, AVIF, JPEG, PNG or GIF, optionally resizing, cropping, rotating, flipping, blurring, sharpening or removing colour. Returns a link valid for 24 hours with the source and output details, and the image itself when it is small enough.
Inputs: url (required), fit, blur, flip, +8 more.
generate_barcodeGenerate a barcodeA
Renders a barcode: Code 128, Code 39, Code 93, EAN-13, EAN-8, UPC-A, UPC-E, ITF-14 or Codabar. Returns a link valid for 24 hours, the dimensions and the image itself.
Inputs: content (required), type, color, scale, +3 more.
generate_qr_codeGenerate a QR codeA
Encodes text or a URL as a QR code. Returns a link to the image and the image itself. With the text format it returns the code as rows of 0 and 1 instead.
Inputs: content (required), size, color, format, +3 more.
generate_social_imageGenerate a social card imageA
Creates an Open Graph image, the card shown when a link is shared, from a title and an optional subtitle, site label and logo. Returns a link valid for seven days and the image itself.
Inputs: title (required), bg, font, logo, +7 more.
get_drive_download_linkGet a Google Drive download linkARead-only
Converts a Google Drive share link into a URL that downloads the file directly.
Inputs: url (required).
inspect_imageInspect an imageARead-only
Reports the width, height, format, file size, orientation, transparency and dominant colours of an image at a public URL.
Inputs: url (required).
inspect_ssl_certificateInspect an SSL certificateARead-only
Connects to a host and reports its TLS certificate: issuer, subject, validity dates, days remaining, whether the chain is trusted and the name matches, and the protocol and cipher in use.
Inputs: host (required), port, fresh.
lookup_dnsLook up DNS recordsBRead-only
Returns the DNS records of a host or domain (A, AAAA, MX, TXT, NS, CNAME, SOA, CAA, SRV) and its DNSSEC status.
Inputs: name (required), type.
lookup_whoisLook up a domain registrationARead-only
Returns the registration record of a domain from RDAP or WHOIS: registrar, creation, update and expiry dates, status codes, nameservers and abuse contact.
Inputs: domain (required), raw.
preview_linkPreview a linkARead-only
Reads the metadata of a public URL: title, description, preview image, favicon, site name, author, publication date, language and feeds. A quick way to learn what a link points to without reading the page.
Inputs: url (required), fresh, render, verify_image.
read_qr_codeRead a QR codeARead-only
Decodes the QR code in an image and returns the text it holds.
Inputs: image (required), format.
read_webpageRead a web page as MarkdownARead-only
Fetches a public web page and returns its main content as Markdown, without navigation, adverts or scripts. For pages that build their content with JavaScript, set render to true.
Inputs: url (required), mode, fresh, links, +3 more.
render_htmlRender HTML to an image or PDFA
Renders an HTML document you provide as a JPEG, PNG or WebP image, or as a PDF: an invoice, a report, a certificate. Returns a link to the file, and the image itself when it is small enough. Images, fonts and stylesheets referenced by the HTML must use absolute URLs.
Inputs: html (required), width, format, height, +1 more.
translate_textTranslate textBRead-only
Translates text from one language to another. 46 languages are supported.
Inputs: to (required), text (required), from.
verify_emailsVerify email addressesA
Checks whether email addresses can receive mail by asking the receiving mail server, without sending a message. Returns valid or invalid for each address. Can take up to a minute.
Inputs: emails (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 19 tools
Each tool targets a clearly distinct operation (QR, DNS, SSL, WHOIS, email, web rendering, image processing, translation). The only mild overlap is among web-content tools (read_webpage, preview_link, capture_screenshot, render_html), but their descriptions clearly differentiate fetching text, metadata, visual capture, and rendering provided HTML.
All 19 tools use snake_case with a leading verb (read_, lookup_, inspect_, check_, verify_, generate_, capture_, render_, convert_, preview_, translate_, get_), creating a highly predictable pattern. No mixing of conventions.
19 tools is on the heavy side for a single MCP server; the rubric treats 16–25 as borderline. Each tool is distinct, but the broad utility scope means the set could feel large to an agent.
The surface covers many common web utilities well: QR/barcode generation and reading, image conversion/inspection, web rendering/screenshots, DNS/SSL/WHOIS lookups, email checks, translation, and Drive link conversion. Minor gaps include OCR beyond QR, PDF-to-text, URL expansion/shortening, or text summarization, but core utilities are present.
Maintenance
Related MCP Connectors
PDF, image, video, OCR, screenshot, SQL, QR and text tools for agents. No API key, no signup.
Read a URL as clean markdown, screenshot a website, url to PDF. Web access for agents, no signup.
A real browser for your agent: screenshots, page checks, Markdown and PDFs.
Web data for agents: YouTube transcripts, screenshots, Google News, WHOIS, jobs, tech stack, more.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables AI assistants to capture webpage screenshots, generate PDFs from URLs or HTML, and extract rich metadata like Open Graph and JSON-LD data. It provides tools for web-to-image/PDF conversion and structured data extraction through the Junipr API.17 npm1MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI agents to read, screenshot, or convert web pages to PDF using a real headless browser, turning any URL into clean Markdown, a visual image, or a print-ready document.82 npmMIT
- AlicenseAqualityBmaintenanceEnables AI assistants to fetch and process web content securely, including HTML-to-markdown conversion, reader mode, metadata extraction, RSS/sitemap parsing, and robots.txt-aware requests with SSRF protection.152,171 npm2MIT
- AlicenseAqualityCmaintenanceEnables AI assistants to access 50+ developer utility APIs for inspecting SSL, DNS, WHOIS, security headers, redirects, sitemaps, and page speed, converting between JSON, CSV, YAML, XML, Markdown, and HTML/PDF, and generating UUIDs, hashes, passwords, QR codes, and screenshots via natural language.5062 npmMIT