hireproof

name

hireproof

description

Investigate suspicious job postings using HireProof's live AI agent tools; use when a user asks you to verify a job listing, check if a company is legitimate, detect recruitment scams, compare salary offers against market rates, or confirm a local business footprint before applying. Also use when the user pastes a job description, recruiter message, WhatsApp screenshot, or offer letter and asks if it is real or safe.

HireProof — Job Investigation Skill

Investigate suspicious job postings end-to-end using live web intelligence. Cross-reference company presence, news reputation, salary benchmarks, and physical footprint. Return a structured verdict (Safe, Caution, or High-Risk) backed by verifiable evidence URLs — never by opinion.

Quick Start

1. User provides a job post (text, URL, or image).
2. Extract claims: company name, role, salary, location, contact method.
3. Call investigation tools concurrently to gather evidence.
4. Score the findings and present a verdict with receipts.

Setup

0) If MCP tools are not available, connect the server first:

MCP tools perform live provider-backed searches. Use an account-issued API key from the Developer Portal with BYOK provider credentials or a private self-hosted API key. Add the server to your MCP configuration and restart:

{
  "mcpServers": {
    "hireproof": {
      "url": "https://hireproof.tech/api/mcp",
      "headers": {
        "x-api-key": "<your-account-api-key>"
      }
    }
  }
}

After adding, restart your client. The 4 live tools will be available for account-issued keys.

Tool naming: Most MCP clients (Claude Code, Cursor, Copilot) namespace tools as server_name:tool_name. Since the server is named hireproof, the tools appear as:

MCP Tool Name	Direct Protocol Name
hireproof:search_company	search_company
hireproof:news_check	news_check
hireproof:jobs_compare	jobs_compare
hireproof:local_presence	local_presence

Use whichever format your client supports. Both resolve to the same tool.

Alternative: REST API (no MCP required)

If your client does not support MCP, use the REST endpoint directly via HTTP:

POST https://hireproof.tech/api/v1/audit
Headers: Content-Type: application/json, x-api-key: <your-account-api-key>
Body: { "text": "<job post text>", "location": "<optional>", "mode": "live" }

This runs the live full pipeline and returns a complete AuditReport as JSON. Deterministic fixture checks still require an account-issued key or private self-hosted API key with "mode": "demo".

Workflow

1) Receive and normalize the input

The user may provide:

• Raw text — job description, recruiter DM, email body, or chat message
• URL — link to a job listing on LinkedIn, Indeed, Jobstreet, etc.
• Image — screenshot of a WhatsApp message, Telegram chat, PDF offer letter, or social media post

Before proceeding, extract these 6 claims from the input:

Claim	Example	If missing
Company name	"Accenture"	Mark as "Unknown / Not stated" — this is itself a red flag
Role / title	"Frontend Intern"	Use whatever description is given
Salary	"PHP 80,000/week"	Note "Salary not disclosed"
Location	"Philippines"	Infer from currency, language, or ask the user
Contact method	"Telegram @recruiter"	Note "Not stated"
Application path	"No interview, start Monday"	Note how the user is expected to apply

Critical rule: If the user provides only an image, describe what you see and extract claims from the visible text. If you cannot read the image, ask the user to paste the text.

2) Choose your investigation strategy

Option A — Full pipeline via REST (recommended for comprehensive results):

curl -X POST https://hireproof.tech/api/v1/audit \
  -H "Content-Type: application/json" \
  -H "x-api-key: <your-account-api-key>" \
  -d '{"text": "<full job post text>", "location": "<location>", "mode": "live"}'

This runs the live pipeline and returns a scored report. For demo fixtures, use your account-issued or private self-hosted API key with "mode": "demo". Skip to Step 4.

Option B — Individual MCP tools (when you need granular control):

Call tools selectively based on what claims are present. Call them concurrently when possible.

Condition	Tools to call
Company name is known	search_company + news_check + local_presence (all 3)
Company name is unknown	jobs_compare only (benchmark salary)
Salary is stated	jobs_compare (always — salary verification is critical)
Location is stated	local_presence + jobs_compare with location
All claims present	Call all 4 tools concurrently

3) Call the tools

search_company — Verify web presence

{
  "method": "tools/call",
  "name": "search_company",
  "arguments": { "company_name": "Accenture", "role": "Frontend Developer" }
}

Returns: Array of evidence items with source, snippet, url, and type. Look for: Official website, LinkedIn company page, Wikipedia entry, Glassdoor profile. Their absence is a strong negative signal.

news_check — Search for scam reports

{
  "method": "tools/call",
  "name": "news_check",
  "arguments": { "company_name": "TechStart Solutions", "keywords": ["scam", "fraud", "fake"] }
}

Returns: News articles with titles, snippets, dates, and URLs. Look for: Fraud warnings, BBB complaints, Reddit threads, government agency alerts. Even one confirmed scam report is a critical red flag.

jobs_compare — Benchmark salary

{
  "method": "tools/call",
  "name": "jobs_compare",
  "arguments": { "role": "Frontend Intern", "location": "Philippines", "level": "Entry Level" }
}

Returns: Comparable job listings with titles, companies, salaries, and URLs. Look for: Is the offered salary 2x+ above the median for similar roles? If yes, flag as "Unrealistically high salary — possible bait."

local_presence — Verify physical footprint

{
  "method": "tools/call",
  "name": "local_presence",
  "arguments": { "company_name": "Accenture", "location": "Philippines" }
}

Returns: Google Maps results with business names, addresses, and ratings. Look for: Verified business listing with reviews. Ghost companies have no map presence.

4) Score and classify

Apply these weighted rules to produce a risk assessment:

Red flags (each adds 10–25 points to risk score):

Flag	Weight	Trigger condition
No verifiable company web presence	+25	search_company returns 0 relevant results
Active scam/fraud reports found	+25	news_check finds confirmed scam articles
Salary 2x+ above market median	+20	jobs_compare shows offered salary is an outlier
No physical business footprint	+15	local_presence returns 0 results
Telegram/WhatsApp/personal email only	+15	Contact method is not corporate email
"No interview" or "start immediately"	+15	Application path skips standard hiring process
Requests money, IDs, or bank details	+20	Post asks for upfront payment or sensitive docs
Vague role description	+10	No specific responsibilities listed
New domain (< 90 days)	+10	Domain registration is recent

Green flags (each subtracts 5–15 points from risk score):

Flag	Weight	Trigger condition
Company has verified LinkedIn page	-10	search_company finds LinkedIn company profile
Official .com/.org domain with history	-10	Domain is established (1+ years)
Company appears on Glassdoor with reviews	-5	search_company finds Glassdoor profile
Salary within market range	-10	jobs_compare shows salary is reasonable
Physical office verified on Maps	-15	local_presence finds verified listing
Standard hiring process described	-5	Post mentions interviews, assessments

Verdict thresholds:

Risk Score	Verdict	What to tell the user
0–39	SAFE	"This appears to be a legitimate opportunity. Standard due diligence still recommended."
40–69	CAUTION	"This listing has some concerning signals. Proceed carefully and verify independently."
70–100	HIGH-RISK	"This listing has multiple red flags consistent with recruitment fraud. Do not share personal information."

5) Present findings to the user

Always use this structure:

## Verdict: [SAFE / CAUTION / HIGH-RISK]
**Risk Score:** [X]/100

### Key Findings
- [Most important finding with evidence URL]
- [Second finding with evidence URL]
- ...

### Red Flags Found
1. [Flag description] — [evidence source and URL]
2. ...

### Green Flags Found
1. [Positive signal] — [evidence source and URL]
2. ...

### Salary Comparison
- Offered: [salary from post]
- Market median: [from jobs_compare results]
- Assessment: [Within range / Significantly above / Suspiciously high]

### Recommended Next Steps
1. [Actionable step based on verdict]
2. [Second step]
3. ...

Presentation rules:

• Never state a verdict without evidence. Every claim must link to a source URL.
• If a tool returned no results, state "No data found" — do not guess or speculate.
• For HIGH-RISK verdicts, always include: "Do not send money, personal identification, or bank details."
• For CAUTION verdicts, suggest specific verification steps the user can take manually.
• For SAFE verdicts, still recommend standard due diligence (research the company, verify the interviewer on LinkedIn).

Error Handling

Situation	Action
MCP tool returns an error	Report which tool failed, present results from tools that succeeded, note the gap in coverage
REST API returns 429 (rate limit)	Tell the user: "Rate limit reached. Try again in 1 minute."
REST API returns 401 (auth error)	Verify the x-api-key header is present and correct
No company name in the input	Skip search_company, news_check, and local_presence. Run jobs_compare for salary verification. Flag "Company name not provided" as a red flag.
Input is in a non-English language	Extract claims as-is, translate the key fields (company, role, salary) for the tools, note the original language
Multiple job posts in one message	Process each separately and present individual verdicts

Example Investigations

Example 1: High-Risk Scam

Input: "Remote frontend intern. PHP 80,000/week. No interview needed. Message us on Telegram @quickhire."

Claims extracted:

• Company: Unknown / Not stated
• Role: Frontend Intern
• Salary: PHP 80,000/week
• Location: Philippines (inferred from PHP currency)
• Contact: Telegram @quickhire
• Application: No interview

Tool results:

1. search_company — skipped (no company name)
2. news_check — skipped (no company name)
3. jobs_compare("Frontend Intern", "Philippines", "Entry Level") — Market range: PHP 15,000–25,000/month
4. local_presence — skipped (no company name)

Scoring:

• No company name stated: +25
• Salary is 12–20x above market rate: +20
• Telegram-only contact: +15
• No interview: +15
• Vague role description: +10
• Total: 85/100 → HIGH-RISK

---

Example 2: Legitimate Opportunity

Input: "Accenture Philippines is hiring a Junior Software Engineer. Competitive salary. Apply via accenture.com/careers. Office in BGC, Taguig."

Claims extracted:

• Company: Accenture Philippines
• Role: Junior Software Engineer
• Salary: Not disclosed ("Competitive")
• Location: BGC, Taguig, Philippines
• Contact: Official website
• Application: accenture.com/careers

Tool results:

1. search_company("Accenture Philippines", "Junior Software Engineer") — Official website, LinkedIn (500k+ followers), Wikipedia, Glassdoor (4.0 rating)
2. news_check("Accenture Philippines") — No scam reports. Recent press about expansion in PH.
3. jobs_compare("Junior Software Engineer", "Philippines") — Market range: PHP 25,000–45,000/month (salary not disclosed, so no comparison possible)
4. local_presence("Accenture Philippines", "BGC Taguig") — Verified office at Net One Center, BGC. Google Maps rating 4.1 with 200+ reviews.

Scoring:

• Verified LinkedIn: -10
• Official domain with history: -10
• Glassdoor profile: -5
• Physical office verified: -15
• Standard hiring process: -5
• Total: max(0, -45) = 0/100 → SAFE

---

Example 3: Caution — Ambiguous

Input: "DataPro Analytics is hiring remote data entry clerks. $25/hour. No experience needed. Send your resume to [e-mail]"

Claims extracted:

• Company: DataPro Analytics
• Role: Data Entry Clerk
• Salary: $25/hour
• Location: Remote (US inferred from USD)
• Contact: [e-mail]
• Application: Email resume

Tool results:

1. search_company("DataPro Analytics") — Found a basic website (datapro-analytics.com), no LinkedIn, no Glassdoor. Domain registered 4 months ago.
2. news_check("DataPro Analytics") — No results (neither positive nor negative)
3. jobs_compare("Data Entry Clerk", "United States", "Entry Level") — Market range: $15–20/hour. Offered salary is slightly above average.
4. local_presence("DataPro Analytics") — No verified business listing found.

Scoring:

• Minimal web presence (no LinkedIn, no Glassdoor): +15
• New domain (< 6 months): +10
• Salary slightly above market: +5
• No physical footprint: +15
• "No experience needed": +10
• Has a corporate email domain (not Gmail): -5
• Has a website: -5
• Total: 45/100 → CAUTION