Skip to main content
Be Bitwise

Writing Effective Bug Reports

Bug Bounty Reporting & Career · Bug Bounty track

5 min read

In this module

The Report Is the Product

Finding a vulnerability is half the job. If you can't communicate the finding clearly, the triage team will misjudge the severity, fail to reproduce, or quietly de-prioritise it. A clean report buys you faster triage, higher bounties, and a stronger reputation. A scrappy one wastes everyone's afternoon — yours included.

Report Structure

A solid bug report follows a consistent structure. Triage analysts read dozens of reports a day; a predictable layout lets them size yours up quickly and accurately.

text
1. Title
2. Severity / CVSS Score
3. Description
4. Steps to Reproduce
5. Impact
6. Proof of Concept (PoC)
7. Suggested Remediation
8. Supporting Evidence (screenshots, videos, logs)

Title: a clear, specific summary of the vulnerability and its impact. "XSS in search parameter" is vague. "Stored XSS in user profile bio field allows session hijacking of any user viewing the profile" tells the reader exactly what the issue is and why they should care.

Description: explain the vulnerability in technical detail. What's the root cause? Which component is affected? What trust boundary is violated? Write for a security engineer who knows the tech stack but has never laid eyes on this particular bug.

Steps to Reproduce: the section that earns or loses the bounty. Provide exact, sequential instructions that let someone else reproduce the issue from a cold start. Include full HTTP requests (exported from Burp), specific URLs, test account credentials if needed, and the exact browser or tool configuration you used.

text
Steps to Reproduce:
1. Log in to https://app.example.com with a standard user account
2. Navigate to Settings > Profile
3. In the "Bio" field, enter: <img src=x onerror=alert(document.domain)>
4. Click "Save Profile"
5. Open a new browser session and log in as a different user
6. Navigate to the first user's profile at /users/{id}
7. Observe that the JavaScript executes in the second user's browser context

CVSS Vector Strings

The Common Vulnerability Scoring System (CVSS) gives you a standardised way to assess severity. Include the CVSS vector string in your report to back up the rating you've picked:

text
CVSS:3.1/AV:N/AC:L/PR:L/UI:R/S:C/C:L/I:L/A:N
Breakdown:
AV:N - Attack Vector: Network (exploitable remotely)
AC:L - Attack Complexity: Low (no special conditions)
PR:L - Privileges Required: Low (needs authenticated account)
UI:R - User Interaction: Required (victim must visit the page)
S:C - Scope: Changed (affects resources beyond the vulnerable component)
C:L - Confidentiality: Low (session token exposure)
I:L - Integrity: Low (can execute actions as victim)
A:N - Availability: None (no availability impact)
Score: 5.4 (Medium)

Bugcrowd uses its own P1-through-P5 taxonomy instead of CVSS. Adjust your severity justification to whichever platform you're reporting on.

CVSS Is a Starting Point, Not a Verdict

CVSS captures the technical characteristics of a vulnerability but does not account for the specific business context. A medium-CVSS XSS on a banking application's transaction page has far greater real-world impact than the same XSS on a marketing landing page. Use CVSS to justify the baseline severity, then supplement it with a business impact statement that explains the real-world consequence.

Screenshots and Video Evidence

Visual evidence removes ambiguity. Include:

  • Screenshots showing the vulnerability triggering (e.g., the XSS alert box with the domain visible)
  • HTTP request/response pairs from Burp Suite showing the exact traffic
  • Video recordings for complex multi-step vulnerabilities or timing-dependent issues
  • Terminal output for server-side vulnerabilities

Annotate every screenshot. Highlight the relevant part of the response, add arrows pointing at the vulnerable parameter, and include captions saying what each image demonstrates. Triagers shouldn't have to squint.

For video proofs of concept, keep them tight. Show the attack from start to finish, no padding, no tangents. Narrate or annotate when the steps aren't obvious from the visuals alone.

What NOT to Include

Knowing what to leave out matters as much as knowing what to include:

  • Don't include excessive data: if you found an IDOR, show one or two records as proof. Don't dump every accessible row.
  • Don't include personal opinions about the organisation's security posture: stick to the technical facts.
  • Don't threaten or imply consequences: phrases like "this could cause millions in damages" without substantiation just torch your credibility.
  • Don't include irrelevant findings: one report per vulnerability, unless you're reporting a chain.
  • Don't include automated scanner output as the entire report: scanners produce false positives and lack context. Verify manually, write a human report.
  • Don't include other users' personal data: redact any personal information you encounter during testing.

The Payoff of Quality Reports

Quality reports correlate directly with better outcomes. Triage teams prioritise clean ones. Organisations pay better bounties when they can immediately see and verify the issue. Your signal score on HackerOne ticks upward with every report accepted without a clarifying back-and-forth. Stack enough of those together and the private invitations start arriving on their own — the same programmes where competition is lower and payouts are higher.