Free hiring tool
Plan a Reference Check
A reference-check plan for an engineering finalist: verify the referees are real, a call script aimed at the risks found in interviews, and a risk register for the answers.
Runs in Claude or ChatGPT
Read-only
Free
What it produces
Give the skill your debrief worries and the referees the candidate named. In about 10 minutes you get:
- 2-4 named risks a referee can confirm or refute
- A verification route for every referee that doesn't rely on contact details from the candidate
- A 15-minute call script: five core questions for everyone, plus targeted questions per risk
- A "do not ask" box for the call
- A risk register to fill in as you call
- The legal steps that match who is calling
- Two short emails: to the candidate and to each referee
Worked example
From the Reference Check playbook this tool uses.
From a run on a fictional startup's request. The founder wanted to ask about sick days and past pay; the skill removed both and explained why.
1. Risks to test
What it needs
| Input | What to give it |
|---|---|
| Candidate summaryoptional | Role, the candidate's CV highlights, and the open risks from interviews |
| Refereesoptional | Who the candidate offered as references (names, companies, relationship) |
Call it from your assistant
In any assistant connected to Ruzora's MCP server, the tool is plan_reference_check. Your assistant fills the arguments from the conversation; this is the shape:
{
"candidate_summary": "..."
}Frequently asked questions
- Is "Plan a Reference Check" free?
- Yes, it is free. Run it in Claude, ChatGPT or any assistant that connects to Ruzora's MCP server, with no API key.
- What does "Plan a Reference Check" need from me?
- Candidate summary (optional): Role, the candidate's CV highlights, and the open risks from interviews. Referees (optional): Who the candidate offered as references (names, companies, relationship).
- How does "Plan a Reference Check" work?
- The tool hands your details and Ruzora's Reference Check playbook to the AI you run it in, which follows the playbook to write the result. Ruzora's server does not run a model for it.
- Does it make the hiring decision?
- This organizes evidence for you. It is not a hiring decision: you make the call.