thinkhard.SI

Prompt repository Hiring & people

Plan a workplace investigation before the first interview

A complaint needs a formal look, and you need a scope, witnesses, questions and a timeline before anyone is interviewed.

Registry no.
PPL-029
Collection
Hiring & people
Laid out for
Claude
Score
10 out of 10
Blanks to fill
6
Added

Hiring & peopleClaude

The prompt

Score 10 of 10, Excellent

Use it when A complaint needs a formal look, and you need a scope, witnesses, questions and a timeline before anyone is interviewed.

  • {{complaint}}
  • {{policies}}
  • {{my_role}}
  • {{people}}
  • {{interim_measures}}
  • {{audience}}
Full prompt and why it works
System prompt
You help HR professionals plan fair, documented workplace investigations. You organize the allegations, the evidence to gather and the interview questions, and you stay neutral. You do not reach findings, judge anyone's credibility or give legal advice.
User message
<complaint>
{{complaint}}
</complaint>

<policies>
{{policies}}
</policies>

<context>
My role: {{my_role}}
People involved, by role or initials: {{people}}
Interim measures already taken: {{interim_measures}}
</context>

Prepare an investigation plan for the complaint above, for {{audience}} to review before any interviews take place. The plan should make the investigation fair to everyone involved and easy to defend later, so tie every step to a specific allegation and policy.

Use these headings:
1. Allegations: each as a separate, neutral statement of what is alleged to have happened, when and where, with the policy clause it may breach.
2. Scope: what the investigation will and will not cover.
3. Evidence to gather: documents, messages, records and logs, for each allegation.
4. Interview order and witnesses, with the reason each person is included.
5. Questions: six to ten open, non-leading questions each for the complainant, the respondent and each witness, tied to the allegations.
6. Safeguards: confidentiality, protection against retaliation, the respondent's chance to answer each allegation, note-taking, and what to do if new allegations come up.
7. Timeline with milestones.

Keep it under 1,000 words. Use neutral wording throughout, such as "alleged" and "the respondent", and no adjectives that judge anyone.

<example>
The form of one allegation, with invented details: "Allegation 2: On or around [date], the respondent made remarks about the complainant's religion during a team meeting (Respect at Work policy, section 3.1)."
</example>

Do not decide whether anything happened, assess credibility, or suggest outcomes or discipline, because those come after the evidence and belong to the decision-maker. Do not invent facts, dates or witnesses that are not in the complaint.

If the complaint involves a senior leader, possible criminal conduct, a safety risk, or someone who has recently raised other concerns, put a line at the top starting "BEFORE STARTING:" that recommends involving employment counsel or an external investigator. If an allegation is too vague to investigate as written, say what needs clarifying with the complainant first.

Variables

{{complaint}}
The complaint as received, or your notes of it. Use roles or initials instead of names if you can.
{{policies}}
The policies the complaint may involve, for example your harassment or conduct policy.
{{my_role}}
For example "HR manager".
{{people}}
The complainant, respondent and possible witnesses, by role or initials.
{{interim_measures}}
Steps already taken, for example a schedule change, or "none".
{{audience}}
Who approves the plan, for example "the HR director and employment counsel".

Why it works

  • Background facts. The complaint, the policies and the context sit in their own XML tags above the task, so Claude keeps what was alleged apart from what the rules say.
  • What to avoid. No findings, credibility calls or suggested discipline, which keeps the plan neutral and defensible before a single interview.
  • If unsure. Senior leaders, possible crimes, safety risks and possible retaliation trigger a BEFORE STARTING line recommending counsel or an outside investigator.

How to use it

  1. Press Fill in variables and type your details into each blank. What you type stays in this browser.
  2. Copy the filled prompt into the AI tool you already use, such as ChatGPT, Claude or Gemini, or open it in the Grader or the Prompt Tester here.
  3. Before you rely on it, try it on three or four real examples. The Prompt Tester runs them all at once and shows which ones go wrong.
Use it from code

Get this prompt as JSON from the free REST API, with your own text in the blanks, or let an AI assistant fetch it through the MCP server. No key.

curl -s https://thinkhard.si/api/v1/prompts/workplace-investigation-plan

With the MCP server added to Claude, ChatGPT, Cursor or VS Code, call get_prompt with {"prompt": "workplace-investigation-plan"}, or just ask for it by name.