AI Finding Generation

AI Finding Generation

The AI Finding Generation feature transforms brief vulnerability titles into comprehensive, professional finding documentation. Save hours of writing while maintaining consistent quality.

How It Works

1. Input: Provide a vulnerability title or CVE ID
2. Processing: AI analyzes the input and generates content
3. Enrichment: If CVE provided, real data is fetched from NVD
4. Output: Complete finding ready for review

Using AI Generation

From New Finding Form

1. Navigate to Findings → + New Finding
2. Enter the finding title
3. Click Generate with AI
4. Wait for generation (typically 5-15 seconds)
5. Review populated fields
6. Edit and save

What Gets Generated

Field	Generated Content
Description	Technical explanation of the vulnerability
Impact	Security and business impact
Remediation	Step-by-step fix instructions
CVSS Score	Suggested severity score
CVSS Vector	Full CVSS 3.1 vector string
CWE ID	Related weakness category
References	Helpful external links

Adding Context

Improve generation quality by providing context:

1. Finding Specifics field: Add details about your specific instance
2. Affected Asset: Mention the target system
3. Evidence Notes: Describe what you observed

Example with context:

Title: SQL Injection
Finding Specifics: Found in the search parameter on /products endpoint
Affected Asset: api.example.com
Evidence Notes: Parameter 'q' vulnerable to UNION-based injection

Generation for Different Finding Types

Known CVEs

For findings with CVE IDs:

1. Enter: CVE-2024-1234 as the title
2. Click Generate
3. AI fetches official CVE data
4. Combines with AI-generated remediation
5. References link to official sources

Generic Vulnerabilities

For common vulnerability types:

1. Enter: Cross-Site Scripting (XSS) in comment field
2. Click Generate
3. AI generates comprehensive template
4. Customize for your specific finding

Custom/Unique Findings

For unique discoveries:

1. Enter descriptive title
2. Add detailed specifics
3. Generate for starting point
4. Heavily customize the output

Two-Pass Generation

NEURO uses a two-pass approach for quality:

Pass 1: Technical Generation

• Core vulnerability description
• Technical impact
• Standard remediation

Pass 2: Evidence Grounding

If you provide evidence notes:

• Integrates your specific observations
• References your evidence
• Tailors recommendations

Template vs. Project Findings

Project Findings

For findings in a specific project:

• Include scope-specific context
• Reference actual evidence
• Link to affected assets
• Project-specific recommendations

Template Generation

For Content Library templates:

• Generic, reusable content
• No project-specific details
• Suitable for any instance
• Standard remediation steps

Toggle “Generate as Template” for library entries.

Editing AI Output

Recommended Edits

After generation, always review and edit:

1. Verify accuracy: Check technical details
2. Add specifics: Include your evidence details
3. Adjust severity: Modify CVSS for your context
4. Customize remediation: Add environment-specific steps
5. Link assets: Associate affected systems

What to Look For

• Outdated information
• Overly generic statements
• Missing context
• Incorrect severity
• Inapplicable remediation

Batch Generation

Generate multiple findings efficiently:

1. Create findings with titles only
2. Open each finding
3. Click Generate with AI on each
4. Review and save

Tip

Start with your most complex findings where AI saves the most time.

Generation Quality Tips

Better Titles = Better Output

Good Title	Why It’s Good
SQL Injection in login authentication	Specific location
Stored XSS via user profile bio	Type and vector specified
CVE-2024-21351	Exact CVE reference
Missing rate limiting on /api/auth	Specific endpoint

Poor Title	Why It’s Poor
SQLi	Too abbreviated
Vulnerability	No context
Issue	Not descriptive
Security problem	Meaningless

Provide Context

More context = better generation:

Title: Insecure Direct Object Reference
Specifics: Changing user_id parameter in /api/orders/{user_id}
           allows access to other users' order history
Asset: api.example.com
Notes: Verified by accessing orders for user_id=1 through 100

Regeneration

If initial output isn’t satisfactory:

1. Add more specific context
2. Click Regenerate
3. Review new output
4. Use best parts from both

Performance

Generation Time

Typical times:

• Simple finding: 5-10 seconds
• CVE lookup: 10-15 seconds
• Complex with context: 15-20 seconds

Factors Affecting Speed

• CVE database lookup
• Amount of context provided
• System load
• Network latency

Troubleshooting

Generation Failed

If generation fails:

1. Check your input isn’t empty
2. Verify network connectivity
3. Try a simpler title
4. Wait and retry

Poor Quality Output

If output quality is low:

1. Be more specific in title
2. Add finding specifics
3. Include evidence notes
4. Try alternative phrasing

---

Next: Learn about CVE Intelligence