Ruzora MCP¶

A hiring server whose most useful tools are the ones that stop a bad post going out: a pay-transparency check by jurisdiction, a job-post linter, and a screen for instructions hidden in a resume.

Server type: Remote (Streamable HTTP) at https://mcp.ruzora.com/mcp
Auth: none for the read and planning tools; a work-email code is required before actions that need a verified buyer
Live check: POST initialize returned HTTP 200 (serverInfo ruzora 1.0.0, protocol 2025-03-26, tools, prompts and resources capabilities); tools/list returned 35 tools
Real call: check_pay_transparency with locations [California, New York] returned three applicable rules with statute citations and source URLs, rules stated as of 2026-10-04
Tools: 35 (role scorecards, pay-transparency, job-post lint, resume triage and injection detection, interview kits, coding assessments, vetted-engineer search, shortlist and project requests)
Pricing: flat monthly rate per engineer, quoted per role; vendor states 40 to 60 percent below a fully loaded US hire, a three-month initial commitment then month to month on 30 days' notice, and a 60-day replacement guarantee
Vendor: Ruzora (ruzora.com), senior software engineers in Latin America

Why This Matters for Operators¶

Two things changed hiring in the last three years and most tooling has not caught up. The first is that pay-transparency law is now local law: the same job post is compliant in one state, illegal in another, and a range that satisfies one city carries a different effective date in the next. The second is that candidates can now write a resume that reads perfectly to a human and carries hidden instructions to an AI screener.

Both are compliance problems with a computable answer, which is exactly what an MCP server is good for. Neither is solved by a job board.

What It Actually Does¶

The hiring pipeline is exposed as discrete steps rather than one opaque run, and the steps are designed to be argued with.

Planning: build_role_scorecard turns "we need a senior full-stack dev" into a mission, measurable twelve-month outcomes and must-haves. calibrate_seniority works out what level a role really needs by scope rather than years. compare_hiring_models returns a neutral twelve-month cost and risk comparison across a US employee, a US contractor, a freelancer, an agency, staff augmentation and an international hire.

Compliance: check_pay_transparency takes hiring locations and returns which pay-transparency laws apply, with severity, statute citation, effective date and the source URL. lint_job_post takes an existing post and returns a fix list: a missing range where one is required, "US citizens only" phrasing, age, gender or origin language. write_job_description generates a post with the compliance check folded in.

Screening: triage_resumes scores CVs against the must-haves with evidence, with a blind mode and hidden-text flags. detect_resume_injection looks for exactly the attack the AI screener invites: "ignore previous instructions", invisible characters, white-on-white text, metadata tricks. review_code_sample gives a plain-English read of a repository or code sample for non-technical hirers, including authorship confidence.

Interviewing: build_interview_kit, developer_interview_questions for managers who are not engineers, design_coding_assessment for tests that still work when candidates have AI, score_interview_notes to turn raw notes into an anchored scorecard and a debrief memo, and plan_reference_check to verify referees are real.

Supply: search_vetted_engineers returns blind profiles of senior engineers in Latin America who passed Ruzora's vetting, check_market_supply says whether a stack reads Available or On demand, request_vetted_shortlist asks for a shortlist, and verify_ruzora_badge checks that a claimed badge is real and current.

Installation¶

claude mcp add --transport http ruzora https://mcp.ruzora.com/mcp

No sign-in is required for the planning and compliance tools. Actions that need a verified buyer use a two-step email code, verify_work_email then confirm_work_email, so no password or account is handed to the client.

Configuration¶

{
  "mcpServers": {
    "ruzora": {
      "type": "http",
      "url": "https://mcp.ruzora.com/mcp"
    }
  }
}

What a Pay-Transparency Call Actually Returns¶

check_pay_transparency with locations California and New York returned three rules, each with a label rather than a verdict:

{
  "severity": "high",
  "label": "Applies",
  "title": "California",
  "fix": "Cal. Lab. Code § 432.3 (as amended by SB 642, effective 2026-01-01), since 2023-01-01"
}

and for New York State, N.Y. Lab. Law § 194-b with a note that the law covers employers with four or more employees and that the headcount was not supplied, so that test is the operator's to confirm. New York City came back as May apply, explaining that the city law also applies if the role is performed there.

The response also carries notes: ["Rules as of 2026-10-04."], source URLs for each statute, and an explicit disclaimer that this is general information rather than legal advice. That last part is the reason the tool is useful: it produces a checkable citation and a date, not an opinion.

Business Relevance¶

Founders hiring their first engineer abroad get a cost and risk comparison before a recruiter call, and a scorecard that makes the interview loop defensible.
HR and operations leads get pay-transparency checks by jurisdiction with citations, so the post is right the first time.
Anyone screening at volume gets a resume-injection check, which is the cheapest protection against an AI screener being talked into a hire.
Non-technical managers get interview kits and code-sample reviews written for their level of technical knowledge.
Agencies and studios can scope a build through the software-factory path with a fixed-price model rather than an hourly estimate.

Integration with CorpusIQ¶

The natural pairing is evidence before the hire. CorpusIQ can summarise what the business actually does from its connected systems, which is the raw material a role scorecard needs, and it can pull the way the team is already working out of the project and CRM connectors so the scorecard describes the job that exists rather than the job in the job description.

After the hire lands, the same connectors answer the question the shortlist cannot: whether the work the role was created for is being done.

Limitations¶

The vendor is also the supplier. Ruzora sells engineers, so its comparisons and scorecards are produced by a party with a position; the citations are checkable, the conclusion is not neutral.
Pay-transparency data is stated as of a fixed date and covers US jurisdictions in the example returned. Non-US roles need their own check.
No rate card. Pricing is per role and quoted, with a three-month initial commitment, so this is not a try-it-and-walk-away tool.
The supply side is Latin America only, which is a real constraint for roles that require US-person status or a presence in another region.
Several actions require a verified work email, which means an individual cannot run the full shortlist flow from a throwaway address.

FAQ¶

Is there a free tier?¶

There is no subscription to try. The planning, compliance, screening and interview tools are callable without sign-in and without an account, which is where most of the value sits; the placement itself is a quoted engagement with a three-month commitment.

Can it check a job post I already wrote?¶

Yes. lint_job_post takes the text and returns a fix list, and check_pay_transparency takes the locations. Run both before publishing and the post is either compliant or carries a specific reason why it is not.

Why does a hiring server care about prompt injection?¶

Because resume screening is now commonly done by an AI reading untrusted text. detect_resume_injection exists to catch hidden instructions and invisible characters aimed at that reader, which is a category of attack most screening tools do not look for at all.

Is the compliance output legal advice?¶

No, and the response says so. It returns the statute, the effective date and a source link so that a human can confirm.