Using Your ATS with Claude Code to Pull Up Detailed Reports

Your applicant tracking system contains useful information about candidates, jobs, placements, and compensation, but pulling that information into a detailed report is not always easy.

AI ATS tools like Stardex reports can show basic numbers, but they may struggle with more specific requests. For example:

• What is the average compensation for a VP of Sales?

• How does compensation vary by location or industry?

• Which candidates are rising VPs versus established executives?

• What is the minimum, maximum, and average compensation for similar profiles?

Claude Code can help you answer these questions using data from your ATS or recruitment database. Instead of manually exporting records, cleaning job titles, and building charts, you can describe the report you need in plain language.

This guide explains how to connect Claude Code to Stardex ATS and use it to create detailed, read-only recruitment reports.

Before You Start

Claude Code can write and run code on your computer, opening up more flexible ways to analyze recruitment data. You can see more examples in our guide to how Claude Code is changing recruiting.

That power comes with added responsibility, so you should understand what Claude plans to do before approving any action. For this guide, we’ll stick to read-only reporting: Claude will retrieve and analyze Stardex data without changing candidate records.

This workflow is designed for teams that have reached the limits of the Stardex MCP connector. For most customers, the MCP connector is still the easiest and most secure place to start.

To follow along, install the Claude Code desktop app first.

What You Can Create

By the end of this guide, you will be able to create:

• A spreadsheet containing actual compensation records from your database

• A table showing the average, minimum, and maximum compensation for a specific job title

• A visual chart showing how compensation is distributed

• Optional AI-generated candidate segments based on seniority, location, and industry

The main compensation report can include:

• Candidate name

• Current job title

• Linked job or job posting

• Total compensation

• Direct Stardex profile link

You can change the job title or reporting criteria whenever you need a different view of your database.

Step 1: Download the Stardex and Claude Code Starter Folder

Start by downloading and unzipping the Stardex + Claude Code starter folder.

Inside the folder, you will find a file named .env. Open this file using any text editor.

If you are using a Mac and cannot see the .env file, open Finder and press:

Cmd + Shift + .

This shortcut shows hidden files.

The .env file is where you will securely add the API key that allows the starter project to retrieve information from Stardex.

Step 2: Create Your Stardex API Key

To create an API key:

1. Open Stardex.

2. Go to Settings → Your Org → Integrations.

3. Select the option to create a new API key.

4. Copy the key.

You can also go directly to:

stardex.com/settings?tab=organization&section=integrations

Only Stardex administrators can create API keys. If you cannot see the Integrations section, ask an admin from your organization to create the key for you.

Return to the .env file and replace the placeholder with your key:

Save the file when you are finished.

Treat your API key like a password. Do not share it publicly, paste it into a public chat, or upload the .env file to a public repository.

Step 3: Open the Folder in Claude Code

Open the Claude Code desktop app. You can also start Claude Code by entering claude in your terminal.

Select Open folder, then choose the starter folder you downloaded and unzipped.

Claude Code can now use the API key stored in the folder to retrieve data from your Stardex account.

Before running anything, tell Claude that you only want it to perform read-only actions. You can use this instruction:

Use this Stardex connection only for read-only reporting. Do not create, edit, or delete any records.

This gives Claude clear limits for the session.

Step 4: Pull Actual Compensation Data from Your ATS

You can now ask Claude Code to retrieve compensation records from your database.

Enter this prompt:

Pull all actual compensation data across my database. For each record, include the person’s name, current job title, job name if it is linked to a job posting, and total compensation.

Claude will then:

• Work through the available records in your database

• Filter the results to include only records where compensation_type = actual

• Organize the requested information into a table

• Save the results as compensation_data.tsv

A TSV file is similar to a CSV file. You can open it in Excel or drag it directly into Google Sheets.

This gives you a clean compensation spreadsheet without having to search for and copy every record manually.

Step 5: Group and Normalize Similar Job Titles

Job titles are rarely written in exactly the same way.

For example, your database may contain all of the following:

• VP of Sales

• Vice President of Sales

• VP, Sales

• VP of Sales – West

• Sales Vice President

These titles may describe similar roles, but a standard report could treat them as separate groups. Claude Code can normalize these variations and analyze them together.

Use a prompt such as:

Group all VP of Sales titles together. Normalize the title variations and show me the count, average, minimum, and maximum compensation. Make a chart.

You can replace “VP of Sales” with any role you want to study, such as:

• Chief Revenue Officer

• Head of Sales

• Account Executive

• Chief Financial Officer

• Head of Marketing

• Software Engineer

Claude will return a table similar to this:

It can also create a compensation distribution chart and save it as a .png file.

This makes it easier to understand the market range instead of relying only on one average figure.

Step 6: Ask More Detailed Questions About the Data

Once Claude has created the main report, you can continue asking questions in plain language.

For example:

Compare VP of Sales compensation in Tier 1 cities with compensation in other locations.

Show the median compensation in addition to the average.

Break down VP of Sales compensation by current industry.

Identify possible outliers and explain why they may be affecting the average.

Create separate charts for base salary, bonus, and total compensation if those fields are available.

You do not need to start the entire process again for every question. Claude can work from the data it has already retrieved during the session.

However, you should rerun the report when you want the latest information from your ATS.

Step 7: Add AI Candidate Evaluation

You can also use Claude to evaluate and segment candidate profiles. This step is optional and requires an Anthropic API key.

Add the Anthropic API key to your .env file:

You can create an Anthropic API key at:

console.anthropic.com/settings/api-keys

Once the key has been added, enter a prompt such as:

Evaluate each VP of Sales profile and segment the candidates by seniority, location, and current industry. Export the results to a CSV file.

Claude can classify each profile using categories such as:

For this example, Tier 1 may include major markets such as San Francisco, New York, Los Angeles, Seattle, Boston, and Chicago.

Claude can export the results to a file such as vp_sales_segments.csv. Each row can also include a direct link to the candidate’s Stardex profile.

An example output may look like this:

Review AI-Generated Categories Carefully

AI evaluation can help you sort a large number of profiles quickly, but the results should not be treated as final facts.

Categories such as “Rising VP” or “Senior VP” depend on the information available in each candidate’s profile and the criteria Claude uses.

Before using these segments in a client report or business decision:

• Review a sample of the profiles manually

• Ask Claude to explain its scoring criteria

• Check that the same criteria were applied consistently

• Correct any unclear or inaccurate classifications

• Avoid using protected personal characteristics in the analysis

You can improve consistency by defining each category in your prompt.

For example:

Classify a candidate as a Senior VP only if the profile shows at least 10 years of leadership experience or responsibility for multiple teams, regions, or business units.

Clear rules make the final report easier to review and defend.

Why Use Claude Code for ATS Reporting?

Instead of being limited to fixed dashboards, you can ask questions based on the project you are currently working on. You can change the job title, location, industry, compensation range, or candidate segment without building a new report manually. Basically, the main advantage of Claude Code is flexibility.

It can also handle several time-consuming parts of recruitment reporting:

• Retrieving records across a large database

• Cleaning inconsistent job titles

• Calculating compensation ranges

• Grouping candidates into useful segments

• Creating spreadsheets and charts

• Repeating the same analysis with updated data

This can be especially useful for executive search firms that need detailed market information for client pitches, compensation discussions, candidate mapping, or internal research.

Tips for Better Claude Code Reports

Be specific about the data you want

Include the exact fields, filters, and output format in your prompt.

Instead of saying "Give me a compensation report",

Try:

Pull actual compensation records for VP of Sales profiles in SaaS companies. Include name, company, location, current title, total compensation, and Stardex profile link. Export the results as a CSV.

Define how titles should be grouped

Tell Claude which title variations should be treated as the same role. You can also ask it to show you the proposed groupings before running the full analysis.

Request both averages and medians

One unusually high compensation record can pull the average upward. Asking for the median gives you another useful view of the typical compensation level.

Ask Claude to show its assumptions

If Claude has to infer a candidate’s seniority, industry, or location tier, ask it to include a short reason for the classification.

Turn Repeated Reports Into Reusable Claude Skills

If you run the same report regularly, you do not have to enter the full instructions every time. You can save the workflow as a Claude Skill and run it again with a simple command.

For more ideas, explore these ready-to-use Claude Skills for recruiters, including workflows for candidate write-ups, job calibration, client meeting preparation, and daily candidate briefs.

Start with the Safest Workflow

Claude Code can turn ATS data into highly detailed, customized reports. However, it is more advanced than a normal software integration because it can write and run code locally.

Keep the first workflow simple:

1. Start with read-only reporting.

2. Pull a limited set of fields.

3. Review the output.

4. Check how Claude grouped the data.

5. Expand the report only after confirming that the results are accurate.

For most teams, the Stardex MCP connector remains the best starting point. Claude Code becomes useful when you need more control, custom calculations, larger exports, or report formats that go beyond the MCP workflow.

With the right limits and clear prompts, you can use Claude Code to turn the information already stored in your ATS into compensation benchmarks, candidate segments, charts, and detailed reports—all without spending hours cleaning spreadsheets by hand.