Writing effective prompts for Burp AI

In Repeater, Burp AI enables you to investigate HTTP traffic using natural language prompts. As with all AI systems, clear and focused prompts produce better quality results.

This guide helps you to write effective prompts that will make your AI-powered security testing more effective.

Be specific and clear

Write prompts that clearly define your testing objective. For example:

These give Burp AI a clearer direction than vague instructions such as Look for vulnerabilities or Check security.

Provide relevant context

Burp AI's effectiveness depends on the context you provide. The more relevant and detailed this context is, the better the results.

Note

Burp AI does not retain a conversation history. Include all necessary context in each prompt.

Give scenario details

Clearly describe the testing scenario in the prompt. This helps Burp AI's analysis to stay focused and accurate. For example:

Highlight relevant sections

Highlighting key areas helps to direct Burp AI's attention to the most relevant parts of a request or response. By emphasizing specific parameters, headers, or unusual elements, you can guide the analysis more effectively.

Be selective, as over-highlighting can dilute Burp AI's focus and reduce the quality of the insights.

Add notes for context

You can include the contents of Repeater's Notes tab when sending a prompt. Do this when the notes add useful information, such as:

Use an iterative workflow

If Burp AI's initial response is too generic or misses key details, refine your approach. You can do this by:

Combine with manual analysis

Burp AI is designed to support, not replace, manual testing. To effectively integrate it into your workflow:

Stay security-focused

Security testing works best when your prompts are rooted in specific, actionable concerns. To get better results:

Examples

The following examples show how well-structured prompts can help Burp AI deliver focused, actionable responses.

API parameter analysis

Prompt: "Examine the 'userId' parameter in this API request. I'm testing for Insecure Direct Object References (IDOR). Analyze the response pattern and suggest tests to verify if users can access other users' data."

Why this works:

Error message investigation

Prompt: "This error response occurred when I submitted malformed JSON. Analyze it for information disclosure issues and classify any exposed details by sensitivity level. Provide recommendations for safer error handling."

Why this works:

Authentication logic testing

Prompt: "I'm testing this login endpoint to determine whether it properly enforces account lockout after repeated failed attempts. Analyze the response behavior and suggest how to confirm whether lockout is functioning correctly."

Why this works: