# Write the check-in once in a skill, then ask everyone

> Claude skills for HR cover your policies and documents. Connect Tracy to interview every employee for check-ins and exit interviews.

- URL: https://tracyinterview.ai/blog/claude-skills-for-hr
- Published: 2026-11-19
- Author: Louis

> **In short** Claude skills for HR are strong at the written work: policy answers, review drafts and onboarding plans in your own wording. Tracy AI Interviewer adds the conversation, interviewing each person on video for check-ins and exit interviews, asking a sharper follow-up when an answer is vague, and returning what each person said with the words behind it. HR reads the results and decides.

## What Claude skills for HR do well

Claude skills for HR keep your policies, tone and layouts in one file that Claude loads when a request calls for it. Anthropic's help center describes skills as folders of instructions, scripts and resources. HR teams use them for policy answers, review drafts and onboarding plans that follow house wording, and Anthropic's own Human Resources plugin starts that work for Cowork users.

A skill can hold the best 30-day check-in questions in the company, but it never sits down with a new hire. That seat needs a connector, and Tracy AI Interviewer connects to Claude to fill it.

## What should HR install first?

Set up in this order, and stop when the job is covered. First, if your team uses Claude Cowork, Anthropic's own Human Resources plugin. Second, use the document skills for the files you produce. Anthropic's PowerPoint, Excel, Word and PDF skills are active on claude.ai whenever you create documents, with no setup, so there is nothing to install. Third, one skill of your own that holds your policies and tone. Fourth, a connector for any work that needs a real conversation.

The order follows what each piece is. Anthropic's help center says skills are folders of instructions, scripts and resources that Claude loads when relevant, and that "MCP connections give Claude access to tools, while skills teach Claude how to use those tools effectively." A skill carries the know-how and a connector carries the reach. The longer comparison is in [MCP vs skills](/blog/mcp-vs-skills).

![Four steps: add the HR plugin, use document skills, write your policy skill, connect Tracy.](/blog/images/claude-skills-for-hr-1.png)

*Where Tracy fits in an HR setup: the plugin, the document skills and your own skill cover the paperwork, then the Tracy connector covers check-ins and exit interviews. Illustration. Source: Anthropic Cowork and skills help pages.*

Paperwork first, conversations last: the connector comes after the plugin and your own skill.

## What is in Anthropic's HR plugin?

Anthropic's Human Resources plugin bundles six skills and six commands for recruiting stages, onboarding, performance reviews, pay analysis and policy questions, and it installs from Cowork on paid plans. Each skill and command, and how to install the plugin, are on [Anthropic's plugin page](https://claude.com/plugins/human-resources).

## Which skills should HR write itself?

The plugin gives you generic structure. Your policies, pay bands and tone are yours, so the most useful HR skill is usually a small one you write. Good first candidates:

1. A policy answer skill: Point it at your handbook text and tell Claude to quote the clause and say when a question needs a person.
2. An onboarding plan skill: Your 30-60-90 layout, your tools, your owner for each week.
3. A review-writing skill: Your competencies and the tone managers should use.

A skill is a `SKILL.md` file with a name and a description. Anthropic's docs cap the name at 64 characters of lowercase letters, numbers and hyphens, cap the description at 1,024 characters, and say the description must state what the skill does and when to use it. Keep employee data out of the skill itself. Anthropic warns that skills with access to sensitive data could be designed to leak it, and tells you to review any skill you download before enabling it, because a malicious skill can direct Claude to call tools or run code against its stated purpose. On Enterprise, owners can turn on content scanning for custom skills uploaded in claude.ai and Cowork.

Where a skill or plugin is enough and where HR needs a connector.

| HR task | Skill or plugin covers | Needs a connector |
| --- | --- | --- |
| Policy answers | Handbook skill quotes your clause | Your document store, if Claude should search it |
| Review draft | `/performance-review` structure and your tone | Nothing; Claude alone |
| Plan onboarding | `/onboarding` and your 30-60-90 layout | Nothing; Claude alone |
| Benchmark pay | `compensation-benchmarking` skill | Your pay data, loaded by you or a connector |
| Check-in | Your question list and summary layout | Tracy: the conversation, on video |
| Exit interview | Your themes and how to report them | Tracy: the conversation, on video |
| Engagement themes | The report layout you want | Tracy: what each person said, with the words behind it |

## The check-in a skill plans but cannot hold

Say your skill writes the check-in list, the follow-up prompts and the layout of the summary. Someone still has to ask the questions. For 100 new hires, a check-in of 20 minutes each is about 33 hours of live calls, and a call that slips to month three no longer answers the 30-day question.

Tracy takes the asking. Each person opens a link and talks with Tracy when it suits them, and the result comes back to the same Claude chat for HR to read. The skill still decides what to ask and how to report it. HR's hours go into reading what people said and acting on it.

## Which connector handles check-ins and exit interviews?

Tracy AI Interviewer handles both. You choose the conversation (an onboarding check-in, engagement or exit), bring your own questions or describe the goal and edit the draft, and send one link. Tracy says who she is and what happens next, and people are told about the recording before it starts.

Asking Claude for results brings back what each person said to each question and the result fields. Grouping answers into themes is done by Claude or the reviewer, not by Tracy. A person reads it and decides what to do. Tracy does not score people or send alerts, and it is SOC 2 Type II attested, with the report on request at hello@tracyinterview.ai. Sample questions are in our [30-60-90 day check-in questions](/blog/30-60-90-day-check-in-questions) and [exit interview questions](/blog/exit-interview-questions). The people teams view is on [the people teams use case](/use-cases/people-teams), and [AI interviews for people teams](/blog/ai-interviews-for-people-teams) goes deeper.

## What Tracy adds to Claude skills for HR

01

### Reach every person

Each person can start on click, at any hour, with no scheduling and no one from your team on the call. Every new hire gets the same check-in.

02

### Ask the next question

Tracy follows what the person said and waits for the answer. A vague answer gets a kinder, sharper follow-up, so HR hears the detail behind a polite reply.

03

### Keep the evidence

Every result includes a recording and a timestamped transcript. Claude reads back what each person said to each question, and HR can open the words behind any answer.

![Four steps: skills and plugins draft documents, Tracy runs the check-in, Claude reads back what was said, HR decides.](/blog/images/claude-skills-for-hr-2.png)

*Tracy runs the check-in on video and Claude reads back what each person said, so HR decides from the words. Illustration.*

Skills draft the documents, Tracy runs the check-in on video, Claude reads back what was said, and HR decides.

## How to connect Tracy to Claude

The steps take a few minutes. On Free, Pro and Max, open Customize, then Connectors, choose Add custom connector, name it Tracy AI Interviewer, and paste `https://tracyinterview.ai/mcp`. Click Connect and Allow. On Team and Enterprise an owner adds it under Organization settings first. The Free plan allows one custom connector. Claude asks for approval before it uses Tracy, and nothing is created or sent until you approve the preview. Every step is in the [Claude connector setup](/blog/tracy-claude-integration).

If your team works in Cowork, Anthropic lists Cowork among the places custom remote connectors work, and says the server must be reachable over the public internet. [A Cowork setup for recruiters](/blog/claude-cowork-for-recruiters) shows the same connector in that app. Pay as you go includes MCP access at $1 per credit, one credit per started 20 minutes, and the first 3 results are free.

## Which path fits you

- HR leads who want the conversation without the calendar: [Tracy for people teams](/use-cases/people-teams).
- Recruiting teams reading the same topic from the hiring side: [Claude skills for recruiting](/blog/claude-skills-for-recruiting).
- Anyone new to the connector idea: [the MCP idea explained](/blog/what-is-an-mcp-server).
- Practical Claude uses beyond skills: [Claude for HR professionals](/blog/claude-for-hr-professionals).

## Common questions

### What are the best Claude skills for HR?

Start with Anthropic's Human Resources plugin if you use Cowork, use the built-in document skills for your files, and write one skill for your own policies. Use skills only from sources you trust.

### Do Claude skills work on every plan?

Anthropic's help center lists skills on Free, Pro, Max, Ultra, Team and Enterprise, with code execution turned on. Plugins are on paid plans.

### Can a Claude skill run an exit interview?

No. A skill holds instructions. Tracy AI Interviewer runs the exit interview on video through the connector and returns what each person said, with the words behind each answer.

### Is it safe to put HR policies in a skill?

Policies and tone are fine. Keep individual employee data out of the file, review anything you download, and check your own data rules before connecting any tool.

### Does Tracy score employees?

No. Tracy reports what people said with a recording and timestamped transcript. A person reads the answers and decides what to do.

### Do people know the conversation is recorded?

Yes. Tracy tells each person on the join screen that the conversation is recorded and transcribed, and says who she is and what happens next, before the interview starts.

---

**Run one check-in from Claude** Connect Tracy to Claude and ask for one 30-day check-in. You approve the draft first. The first 3 results are free. [Try it free](https://tracyinterview.ai/connect?utm_source=blog&utm_medium=seo&utm_campaign=claude-skills-for-hr&utm_content=try-free) [Book a demo](https://cal.com/tracy-ai/tracy-demo-web?utm_source=blog&utm_medium=seo&utm_campaign=claude-skills-for-hr&utm_content=demo)

## Sources

- [Anthropic: Human Resources plugin](https://claude.com/plugins/human-resources), retrieved 14 Nov 2026.
- [Anthropic knowledge-work plugins: human-resources](https://github.com/anthropics/knowledge-work-plugins/tree/main/human-resources), retrieved 14 Nov 2026.
- [Anthropic help center: What are skills?](https://support.claude.com/en/articles/12512176-what-are-skills), retrieved 14 Nov 2026.
- [Anthropic: Agent Skills overview](https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview), retrieved 15 Nov 2026.
- [Anthropic help center: Use plugins in Claude Cowork](https://support.claude.com/en/articles/13837440-use-plugins-in-claude-cowork), retrieved 14 Nov 2026.
- [Add Tracy AI Interviewer to Claude (public guide)](https://tracyinterview.ai/help/claude), retrieved 14 Nov 2026.
