JobVerify
Allows checking the history and content of websites via the Internet Archive's Wayback Machine, providing access to archived snapshots to verify how long a page or domain has existed.
You get a message: "We loved your profile and want to offer you a great remote job." It sounds real. The company has a logo. The recruiter has a photo. But something feels off.
JobVerify helps you find out — in seconds — whether it's genuine or a scam.
You paste the recruiter's message (or a company name, a link, or an email) and ask your AI assistant. JobVerify quietly runs the same background checks a professional investigator would, then gives you a plain-English answer: looks legit, be careful, or this is almost certainly a scam — and why.
❌ Without JobVerify
Fake recruiter and job-offer scams are everywhere, and they're convincing. On your own, you're left guessing:
❌ Is this company real, or a name someone invented last week?
❌ Is this "recruiter" a real person, or a stolen photo and a throwaway account?
❌ Is that application link safe, or a look-alike site built to steal your data?
❌ Why are they asking me to pay for equipment, or move the chat to WhatsApp/Telegram?
By the time you notice, your time, your personal details — or your money — may already be gone.
Related MCP server: openosint
✅ With JobVerify
JobVerify cross-checks the offer against dozens of free, public information sources and combines the clues into one clear verdict:
✅ Tells you if the company is a registered, real business — or nowhere to be found
✅ Spots classic scam scripts (upfront fees, fake "task" jobs, crypto, "let's move off-platform")
✅ Flags suspicious links, look-alike domains, and brand-new websites made to look official
✅ Checks whether the email, phone, and photos actually belong to who they claim
✅ Explains its reasoning in everyday language, so you can decide with confidence
No account, no sign-up, and none of your data is sold or stored. It simply helps you not get fooled.
🕵️ How it works
You don't need to learn anything technical. It's a three-step conversation:
Paste it. Drop the recruiter's message, a company name, a job link, or an email into your AI chat.
It investigates. JobVerify pulls out every detail — the company, links, email, phone, wallet addresses — and quietly checks each one against public records, scam databases, and website history.
You get a verdict. A short, honest summary: how risky it looks, which signals are reassuring, which are red flags, and what to do next.
JobVerifynever logs into or scrapes LinkedIn. It only looks at information that is already public, and reads website history through the Internet Archive — the safe, legal way to check how long a profile or company page has really existed (scammers rely on brand-new throwaway accounts).
🔎 What it looks at
Think of it as a checklist a careful friend — who happens to be a fraud investigator — would run for you:
Area | The question it answers |
The message | Does this match known scam playbooks (advance fees, fake tasks, crypto, urgency)? |
The company | Is it a real, registered business? Any recent scam reports? Is the office address real? |
The person | Is the email real and deliverable? Is the phone valid? Are the photos/usernames reused elsewhere? |
The links | Is the domain brand-new? A look-alike of a real brand? On any phishing/malware blocklist? |
The money | Is the crypto wallet they gave you already flagged in scam databases? |
The history | How long has this profile or website actually been online? |
💬 Example
Ask your assistant something as simple as:
Is this recruiter legit?
"Hi! I'm a talent partner at Example Corp. We loved your profile and
want to offer you a remote role at $45/hr. To get started, please purchase
$200 of onboarding equipment through this link — you'll be fully reimbursed
on day one. Let's continue on Telegram: @examplecorp_hr"JobVerify will pick out Example Corp, the link, and the Telegram hand-off, check each one, and reply with something like: "⚠️ High risk — the company has no public registration, the link was registered 4 days ago, and asking you to pay upfront and move to Telegram are textbook scam signals."
🚀 Getting started
No cloning. No virtualenv. No manual install. JobVerify runs straight from GitHub — you only paste a few lines into your AI assistant's config, and it fetches and launches itself on demand.
The only thing you need once isuv — a tiny, free helper that runs the tool for you:
macOS / Linux:
curl -LsSf https://astral.sh/uv/install.sh | shWindows:
powershell -c "irm https://astral.sh/uv/install.ps1 | iex"
Claude Desktop
Add this to your claude_desktop_config.json, then restart Claude:
{
"mcpServers": {
"jobverify": {
"command": "uvx",
"args": ["jobverify-mcp"]
}
}
}Claude Code
One line in your terminal:
claude mcp add jobverify -- uvx jobverify-mcpThe first run takes a few seconds while it downloads the tool; after that it's instant.
Use it
Paste a suspicious message and ask "Is this offer legit?" — or run the built-in analyze prompt.
That's it.
🔐 Privacy & honesty
Your data stays yours. No sign-up, no tracking, nothing you paste is stored or sold.
No API keys or costs. Every source is free and open — public business registries, DNS records, scam blocklists, the Internet Archive, and more.
Signals, not certainty. JobVerify gives you strong decision support, not a courtroom verdict. Treat the result as informed guidance about a message — never as a final judgment about a real person.
Under the hood, the AI assistant orchestrates these individual checks (all free, no keys):
Tool | What it checks |
| Pulls emails, links, phones, wallets, and profile URLs out of a message |
| Matches text against known scam tactics |
| Email deliverability + linked social accounts |
| Domain age, registrar, and whether it can be spoofed |
| Look-alike / imposter domains |
| Phishing & malware blocklists |
| A site's certificate & subdomain history |
| Origin, SPF/DKIM/DMARC, and mismatches in raw email headers |
| Phone number validity and region |
| Known-scam crypto wallet databases |
| Business registration, press, and real address |
| Whether an org/username really exists and how old it is |
| Internet Archive history & content (the legal way to read a page) |
There's also a single analyze prompt that runs the whole extract → check → verdict flow for you.
⚠️ Disclaimer
JobVerify is a decision-support tool. Its signals are probabilistic and may be incomplete or wrong. Always use your own judgment, and never treat a result as a definitive statement about any individual or organization.
License
MIT — use it, share it, and help people stay safe.
Available Tools
21 toolscheck_certificate_transparencyARead-onlyIdempotent
Look up a domain's SSL certificate history via crt.sh (no key). A cert first seen only days ago = fresh phishing infrastructure; a long history and many subdomains = established. Complements check_domain (RDAP age).
Use when: you have a company/link domain and want infra age corroboration.
Inputs: params (required). Returns: result
check_crypto_addressARead-onlyIdempotent
Check a BTC/EVM crypto address against open scam databases (no key). Any request to pay or receive crypto for a job is itself a major red flag; a listed address is a strong fraud signal.
Use when: extract_entities found a crypto address in the message.
Inputs: params (required). Returns: result
check_domainARead-onlyIdempotent
Look up domain registration via RDAP and assess its age. Freshly registered domains posing as established companies are a strong scam signal.
Use when: you have a company website, job-portal link, or email domain.
Inputs: params (required). Returns: result
check_domain_authARead-onlyIdempotent
Check a domain's SPF and DMARC DNS records (no key). A domain with no SPF and weak/absent DMARC (p=none) is trivially spoofable — so a 'recruiter' email from it is easy to fake. Strong DMARC + SPF means hard to impersonate.
Use when: you have the sender's / company's domain.
Inputs: params (required). Returns: result
check_emailARead-onlyIdempotent
Assess a recruiter's email: disposable domain, free provider, MX records, and (if given) whether it matches the claimed company's domain.
Use when: you have the sender's email address. Returns a risk-scored result (red/yellow/green) with findings.
Inputs: params (required). Returns: result
check_email_footprintARead-onlyIdempotent
Check an email's public Gravatar profile & linked social accounts (no key). An established identity is a mild legitimacy signal; a throwaway scam address usually has none. Corroborate the linked accounts against the recruiter.
Use when: you want a quick digital-footprint read on a sender's email.
Inputs: params (required). Returns: result
check_github_orgARead-onlyIdempotent
Verify a company's GitHub organization (no key): whether it exists, its age, repo count, and whether its website matches the company domain. A real tech company usually has a GitHub org linking back to its site; a scam rarely does.
Use when: the company claims to build software/tech.
Inputs: params (required). Returns: result
check_ipARead-onlyIdempotent
Check an IP against local blocklists (FireHOL abuse aggregate + Tor exit list) and identify its ASN/org/country (iptoasn) with a hosting/datacenter heuristic. No API key. Blocklisted or datacenter IPs are red flags.
Use when: you have an originating IP (e.g. from email headers).
Inputs: params (required). Returns: result
check_phoneARead-onlyIdempotent
Validate a phone number (offline, libphonenumber): validity, type (mobile/VoIP/premium), region, carrier. VoIP/invalid numbers are red flags.
Use when: a recruiter gives a phone/WhatsApp number.
Inputs: params (required). Returns: result
check_scam_patternsARead-onlyIdempotent
Scan message text for known recruiter/job-scam tactics (offline, no key): advance fee, fake check, equipment purchase, task scam, reshipping, crypto payment, personal-docs-early, off-platform push, urgency, no-interview offer.
Use when: you have the raw offer text and want deterministic TTP hits.
Inputs: params (required). Returns: result
check_typosquattingARead-onlyIdempotent
Detect lookalike / typosquatting / homograph domains (offline, no key): brand embedded with extra words ('google-careers'), near-miss typos ('linkedln'), IDN/punycode confusables, and high-abuse TLDs ('.top').
Use when: you have a domain and a real brand it might be impersonating.
Inputs: params (required). Returns: result
check_urlARead-onlyIdempotent
Check a URL/domain against local phishing + malware blocklists — no API key. Uses Phishing.Database (~390k phishing domains) and the URLhaus recent feed (malware URLs). Any hit means high risk.
Use when: the offer contains a link (application portal, form, download).
Inputs: params (required). Returns: result
check_usernameARead-onlyIdempotent
Check a username/handle across GitHub, Reddit and Keybase (no key). Reveals online history and (via GitHub) account age. A handle that exists nowhere is a sock-puppet signal. For broader coverage, also web-search the username.
Use when: you have a recruiter's handle/username to vet.
Inputs: params (required). Returns: result
check_waybackARead-onlyIdempotent
Check the Internet Archive history of a URL (e.g. a LinkedIn profile or company site). A missing or very recent first snapshot suggests a new page. This is the legal proxy for 'account age' since LinkedIn hides creation dates.
Use when: you have a profile/company URL and want an age lower-bound.
Inputs: params (required). Returns: result
extract_entitiesARead-onlyIdempotent
Parse a raw recruiter message into structured entities (offline, no key): emails, domains, URLs, LinkedIn URLs, phone numbers, IPs, crypto addresses, and off-platform (WhatsApp/Telegram) mentions. Run this FIRST, then feed each entity to the specific check_* tools so nothing is missed.
Use when: you have a raw message/offer and want the entities to investigate.
Inputs: params (required). Returns: result
fetch_archived_pageARead-onlyIdempotent
Fetch the archived TEXT of a page from the Internet Archive (no key). This is the legal way to read a LinkedIn profile/company page — it reads the web.archive.org snapshot, NOT LinkedIn live. Returns the extracted text plus the snapshot date so you can evaluate the recruiter's headline/company or a company page's about text.
Use when: you have a LinkedIn (or company) URL and want to evaluate its content.
Inputs: params (required). Returns: result
find_lookalike_domainsARead-onlyIdempotent
Generate typo/homoglyph/TLD permutations of a REAL brand domain and return the ones that actually resolve in DNS (no key). Live look-alikes are prime fake-recruiter / phishing infrastructure.
Use when: you know the real company domain and want to hunt impersonators.
Inputs: params (required). Returns: result
parse_email_headersARead-onlyIdempotent
Analyze raw email headers for spoofing (offline, no key): SPF/DKIM/DMARC results, From vs Return-Path vs Reply-To mismatches (reply-hijacking), and the real originating IP (feed it to check_ip).
Use when: the user can paste the full raw headers of a suspicious email.
Inputs: params (required). Returns: result
search_company_newsARead-onlyIdempotent
Search regional/local news for a company via Google News RSS (no key). Especially useful for SMALL companies that GLEIF/SEC don't cover: a real firm usually has some press footprint in its region, and any headline flagging it as a scam/fraud is decisive.
IMPORTANT: determine the company's ACTUAL country FIRST (from its stated HQ
address, its website's ccTLD, its GLEIF jurisdiction, or a web search) and
pass it as country (or pass company_domain to infer it). Do NOT rely on a
default region — searching the wrong country hides real local coverage and a
"no news" result then means nothing.
Use when: verifying a small/local company beyond registries.
Inputs: params (required). Returns: result
verify_addressARead-onlyIdempotent
Geocode a physical address via OpenStreetMap (no key) and assess whether it resolves and is a business vs residential location. A company 'HQ' that does not resolve, or resolves to a house, is a red flag.
Use when: an offer lists a company address to verify.
Inputs: params (required). Returns: result
verify_companyARead-onlyIdempotent
Search the global LEI database (GLEIF, no API key) for a company name. A match with ACTIVE status is strong evidence the entity is real. Note that only entities with an LEI are listed, so 'no match' is weak evidence.
Use when: a company name is claimed in the offer.
Inputs: params (required). Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 21 tools
Each tool has a clearly distinct scope (domain age, email auth, phone validity, crypto scam DB, etc.), with explicit usage guidance. No two tools overlap in purpose.
Most tools follow a 'check_' or 'verify_' prefix pattern, with a few exceptions like 'extract_entities' and 'parse_email_headers'. All use snake_case and descriptive verbs, making the set mostly consistent.
21 tools is above the typical 3-15 range but justified by the comprehensive domain coverage. Each tool addresses a distinct verification need, so the count feels well-scoped rather than bloated.
The tool set covers a wide range of signals (domain, email, phone, crypto, username, company, URL, headers, entities). Minor omissions like broader social media checks are acceptable given the focused use case.
Maintenance
Related MCP Connectors
A hosted LinkedIn MCP server for your own account. Paste one address into Claude, ChatGPT, Cursor or n8n to read your inbox, send messages and invitations, and search people. Free to start. Not made by LinkedIn.
Free WhatsApp MCP server for ChatGPT and Claude: read and search chats, every send you approve.
MCP server giving Claude AI access to 22+ NYC public-record databases for real estate due diligence
Managed LinkedIn MCP server for AI agents: search, connect, message and enrich on accounts you own.
Related MCP Servers
- AlicenseAqualityAmaintenanceCyberSecurity MCP Server extends Claude with real-time cybersecurity reconnaissance capabilities that Claude doesn't have by default. Instead of manually running 5 different tools across different terminals, just tell Claude "analyze google.com" and get a complete security breakdown instantly. Tools included: * WHOIS Lookup — registrar, ownership, creation/expiry dates * DNS Enumeration — A,828MIT
- AlicenseAqualityBmaintenanceAI-powered OSINT framework exposing 10 tools (email, username, breach, WHOIS, IP, subdomain, phone, Shodan, dorks, Pastebin) as an MCP server for Claude Code and Claude Desktop, with an autonomous agent REPL and direct CLI.2201,221 PyPI1,938MIT
- AlicenseNot gradedqualityDmaintenanceA comprehensive MCP server that exposes multiple OSINT tools to AI assistants like Claude, enabling sophisticated reconnaissance and information gathering tasks using industry-standard OSINT tools.250MIT
- AlicenseAqualityDmaintenanceAn MCP server that lets AI assistants like Claude read LinkedIn data through your own logged-in browser session. Access profiles and companies, search for jobs, or get job details.19Apache 2.0