How to use the IDOR Testing Checklist agent skill

What the skill makes the agent do

An agent skill is a folder containing instructions that an AI agent loads when the task matches. This one is built around a single file, SKILL.md, which holds an IDOR (Insecure Direct Object Reference) methodology. Once the skill is active, the agent treats that methodology as its operating checklist and follows the steps in order unless you say otherwise.

The steps are: create two test accounts (one attacker, one victim), discover features that return sensitive data or change it, intercept the traffic with a proxy, and swap identifiers in the sensitive requests. If the first attempts fail, the checklist covers protection bypass techniques, such as decoding encoded IDs, supplying an ID where the application does not ask for one, or changing the request method or file type. The agent tracks which items are done and suggests next steps based on what it finds.

The file is a text methodology only. It contains no scripts and does not make network requests.

How to install the offensive-idor skill

You can install the skill in either of two ways. The first is to download the ZIP from this page and unzip it into your personal skills folder, ~/.claude/skills/. The second is to unzip it into a project's .claude/skills/ folder so that it applies only to that project.

The second way is to upload the skill in the Claude apps under Settings → Capabilities → Skills.

• Personal install: download the ZIP from this page and unzip it into ~/.claude/skills/
• Project install: unzip it into the .claude/skills/ folder of your project
• Claude apps: upload the skill under Settings → Capabilities → Skills

How to trigger the skill with natural requests

You do not need a special command. The skill is meant to activate when your conversation involves IDOR or related terms. The source lists these trigger phrases: IDOR, insecure direct object reference, horizontal privilege escalation, vertical privilege escalation, object enumeration, GUID, API IDOR, mass assignment, and broken access control.

Describe the job in plain terms and include the scope. Saying which accounts you control and what the target is helps the agent apply the checklist to your situation.

• Test this web app for IDOR. I have two accounts I created for the engagement, and the scope is the billing and documents pages.
• Check whether this API lets a regular user read another user's orders by changing the order ID.
• Walk me through horizontal privilege escalation testing on the profile endpoints.
• What should I try next, given that changing the file ID returned a 403?

What you need before you start

The skill tells the agent what to do but does not supply the tools. You need a web proxy to capture and modify requests. The file recommends Burp Suite or OWASP ZAP. The agent does not install either one, so you must set one up yourself.

Arjun is optional. The file mentions it for parameter discovery, which helps when a request does not yet contain an ID parameter. For mobile testing, the file says to install the proxy's CA certificate on the device or emulator so HTTPS traffic can be intercepted. It names mitmproxy and Burp Mobile Assistant as examples.

• A web proxy such as Burp Suite or OWASP ZAP (recommended in the file, not installed by the agent)
• Optional: Arjun for parameter discovery
• For mobile testing: a proxy CA certificate installed on the device or emulator

A worked example of the checklist in action

Suppose you are authorized to test an application and you have two accounts, one for an attacker role and one for a victim role. You ask the agent to test the messages feature. The checklist has the agent look for identifiers in URLs, request bodies, cookies, and headers, and to check common parameter names such as user_id, account_id, and uuid.

The file also suggests adding a missing ID to a request that does not contain one. For example, the source gives the pattern of changing a plain messages request to include a victim's identifier, shown below. If the server returns the victim's data to the attacker account, that indicates an IDOR and the agent will help you document it. If the server refuses, the checklist moves on to bypass techniques and information-leak monitoring.

Safety notes: scripts, network and authorization

The skill itself runs no scripts and uses no network access. All live traffic comes from the proxy and the target you choose to test, so the safety of the engagement depends on your authorization.

Only test systems you own or have written permission to test. The source warns that switching IDs on other people's accounts can be illegal and can cause real harm. Keep your test accounts separate from real user data, and stay within the scope you have been given.

• The skill runs no scripts and makes no network requests itself
• Use it only on targets you own or are explicitly permitted to test
• Do not switch IDs on other people's real accounts

Tips for getting useful results

Give the agent a clear picture of your setup: which accounts are attacker and victim, what roles they hold, and what the proxy has captured. The checklist asks the agent to track progress, so a clear starting point makes its suggestions more precise.

Work through the steps in order, as the skill recommends, and record each request you change. The file also points to newly added features, mobile applications, and older parts of an application as places where access controls may be weaker.

• Create at least two accounts in the same role, and one with a different privilege level if the app has one
• Capture traffic while you use the sensitive features before you change anything
• Ask the agent to keep a list of checklist items you have completed