Connect VirusTotal to Claude via MCP: Threat Reports

💡 No install? Use cloud.anythingmcp.com directly. Sign in, click Connectors → VirusTotal, paste your credentials, mint an MCP API key and you're done. No Docker, no git clone, no local server.

VirusTotal + Claude

Threat intelligence from the VirusTotal API v3: file reports by hash, URL, domain and IP reports from about 70 engines, search, relationships between indicators, sandbox behaviour and MITRE ATT&CK, community comments and votes, rescans and, once an admin enables it, URL scanning.

Prerequisites

See the full setup instructions baked into the connector (visible in the in-app store when you select the connector). The required environment variables for this connector are:

VIRUSTOTAL_API_KEY

Step 1 — Get credentials

Setting up takes about 2 minutes. You need a VirusTotal account; the community API key is free.

1. Sign up for the free VirusTotal Community at https://www.virustotal.com/gui/join-us and confirm the e-mail.
2. Open your API key at https://www.virustotal.com/gui/my-apikey and paste it into VIRUSTOTAL_API_KEY. It is 64 lowercase hexadecimal characters and is sent in the x-apikey header.
3. Install the connector and run virustotal_list_popular_threat_categories: a cheap call that confirms the key.

Public or Premium key: a free key allows 4 requests a minute and 500 a day, and the VirusTotal terms forbid using the public API in commercial products or business workflows. Premium keys have no such limits and unlock virustotal_intelligence_search, more relationships and more context. Daily quotas reset at 00:00 UTC. virustotal_get_api_usage with your VirusTotal user name shows the daily usage per endpoint for up to 30 days.

Privacy: every indicator you look up or submit becomes part of the VirusTotal dataset and is visible to the community. Never send URLs, domains or hashes that are confidential or contain personal data, such as a URL with a token, a password reset link or an internal host name.

What it reads:

• Verdicts: virustotal_get_file_report (by SHA-256, SHA-1 or MD5, no upload), virustotal_get_url_report, virustotal_get_domain_report and virustotal_get_ip_report, with the detection stats of about 70 engines and URL blocklists, each engine's verdict, reputation, votes, tags and, depending on the object, file type and names, final URL and redirects, DNS records, WHOIS, registrar, ASN and country.
• virustotal_search takes a hash, a full URL, a domain, an IP or a comment tag such as #phishing. With a URL it returns the URL report, the easy way when you do not have the URL id.
• Relationships: virustotal_list_related_objects follows one, for example a file's contacted_domains, a domain's resolutions or subdomains, an IP's communicating_files. virustotal_list_related_ids returns only the ids, which is much smaller. virustotal_get_resolution reads one DNS resolution (id = the IP followed by the domain, for example 8.8.8.8dns.google).
• Sandboxes: virustotal_get_file_behaviour_summary (processes, files, registry keys, mutexes, DNS, IP and HTTP traffic) and virustotal_get_file_mitre_attack (MITRE ATT&CK tactics and techniques per sandbox).
• Community: virustotal_list_comments and virustotal_list_votes on one object, virustotal_list_latest_comments across VirusTotal (filter such as tag:malware), virustotal_get_comment for one.
• Premium only: virustotal_intelligence_search with search modifiers such as type:peexe positives:10+.

What it writes: virustotal_rescan asks for a fresh analysis of a file, URL, domain or IP VirusTotal already knows, and virustotal_get_analysis reads the result (poll every 15 to 60 seconds until status is completed). virustotal_add_comment posts a public comment under your account (words starting with # become tags) and virustotal_add_vote casts a public harmless or malicious vote.

Switched off at install: virustotal_scan_url and virustotal_delete_comment. Neither can be undone: a submitted URL, with anything in it, is shared with the VirusTotal community for good, and a deleted comment is gone. A workspace admin turns them on when the team needs them. With virustotal_scan_url enabled, a URL VirusTotal has not seen (404) can be submitted and its result read with virustotal_get_analysis.

URL identifiers: URL tools take the URL id, which is the URL in unpadded base64url (https://example.com/ becomes aHR0cHM6Ly9leGFtcGxlLmNvbS8) or the SHA-256 of the canonical URL. The data.id of a URL report is that SHA-256; an analysis id looks like u-<sha256>-<timestamp>. When in doubt, the AI uses virustotal_search with the URL.

Data formats: answers follow JSON:API with data.id, data.type and data.attributes. last_analysis_stats counts the engines per verdict (malicious, suspicious, harmless, undetected, timeout), last_analysis_results lists each engine's category and result. reputation is a community score, negative is bad; total_votes counts harmless and malicious votes. Dates are Unix timestamps in seconds (UTC). A low detection count is no proof that something is safe: the AI also checks the analysis date and the relationships. URL, domain and IP reports leave out bulky fields such as HTTP headers, cookies, HTML meta and the full HTTPS certificate; certificates are available through the historical_ssl_certificates relationship.

Paging: list tools return meta.cursor and links.next while more items exist; the AI passes meta.cursor as cursor with the same limit and stops when it is missing.

Common errors: 401 WrongCredentialsError or AuthenticationRequiredError means a wrong or missing key (or an unconfirmed account e-mail), 403 ForbiddenError a Premium-only endpoint or relationship, 404 NotFoundError an object VirusTotal has not seen, 409 AlreadyExistsError a duplicate vote, 429 QuotaExceededError the minute or daily quota (wait, or upgrade), 400 InvalidArgumentError a malformed id.

The connector does not upload files or download samples: look files up by hash.

Step 2 — Install the adapter

curl -fsSL https://raw.githubusercontent.com/HelpCode-ai/anythingmcp/main/docker-compose.quickstart.yml -o docker-compose.yml
printf 'JWT_SECRET=%s\nENCRYPTION_KEY=%s\n' "$(openssl rand -hex 32)" "$(openssl rand -hex 32)" > .env
docker compose up -d

Step 3 — Add the connector in Claude

Your server URL: in AnythingMCP, open MCP Servers → the server this connector is on, and copy its URL (https://cloud.anythingmcp.com/mcp/…). Use it wherever this guide shows YOUR_SERVER_ID.

Recommended path: works on claude.ai web without editing any config file.

1. In Claude, open Customize → Connectors.
2. Click +, then Add custom connector.
3. Enter a name, e.g. VirusTotal, and the remote MCP server URL https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID, then click Add.
4. Click Connect and sign in to AnythingMCP (OAuth) to allow access.

All tools below appear in your chat: start typing prompts.

Available tools

Example prompts

• "Is the file with SHA-256 275a021bbfb6489e54d471899f7db9d1663fc695ec2fe2a2c4538aabf651fd0f flagged as malicious? Show the detection stats and the threat label."
• "Check the domain example-login-secure.com: what do the engines say, when was it created and which IPs did it resolve to?"
• "What is the reputation of the IP 185.220.101.1, who owns it and which files have communicated with it?"
• "Show the sandbox behaviour and the MITRE ATT&CK techniques of this hash: 44d88612fea8a8f36de82e1278abb02f."
• "List the latest community comments tagged phishing and summarise which brands they target."

FAQ

Does it work with Claude Code as well as Claude Desktop? Yes, point both at https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID.

Next steps

• Connector landing page
• Connect to ChatGPT
• Connect to Cursor
• Connect to Gemini
• Connect to Meta Muse
• Connect to Microsoft Copilot