triage-findings

name	triage-findings
description	Deduplicate, prioritize, and rank vulnerability findings using CVSS, EPSS, and business context. Use when processing scan results, correlating multiple findings, determining remediation priority, or when the agent needs to organize discovered vulnerabilities.
tags	["security","triage","cvss","prioritization","deduplication"]
triggers	["triage findings","prioritize vulnerabilities","deduplicate findings","rank vulnerabilities"]

triage-findings

When to Use

• Processing vulnerability scan results that need prioritization
• Multiple findings need deduplication and correlation
• Determining which vulnerabilities to report first
• Correlating findings across different tools or endpoints
• Creating actionable remediation priority lists
• Filtering false positives from valid findings
• Assessing business impact of discovered vulnerabilities

Quick Start

Basic triage workflow:

## Findings Triage Summary

### Critical Priority (Immediate Action Required)

| CVE            | CVSS | EPSS  | KEV | Finding   | Affected |
| -------------- | ---- | ----- | --- | --------- | -------- |
| CVE-2024-XXXXX | 9.8  | 87.5% | Yes | RCE in... | server01 |

### High Priority

| CVE            | CVSS | EPSS  | KEV | Finding    | Affected        |
| -------------- | ---- | ----- | --- | ---------- | --------------- |
| CVE-2024-YYYYY | 8.1  | 23.4% | No  | SQLi in... | app.example.com |

### Deduplicated (Same Root Cause)

- CVE-2024-ZZZZZ: Found on 5 endpoints (single patch needed)

### False Positives Identified

- Finding X: Version mismatch (patched version detected)

Step-by-Step Process

Step 1: Collect All Findings

Aggregate from multiple sources:

Sources to consolidate:

- Automated scanner results (Nuclei, Burp, etc.)
- Manual testing findings
- Exploit agent confirmed vulnerabilities
- Recon-discovered exposures

Capture for each finding:

1. Vulnerability type/name
2. CVE ID (if applicable)
3. Affected endpoint/asset
4. Evidence/proof
5. Tool that found it
6. Timestamp discovered

Initial Data Structure:

{
  "findings": [
    {
      "id": "FIND-001",
      "cve": "CVE-2024-12345",
      "type": "SQL Injection",
      "endpoint": "https://example.com/api/search",
      "parameter": "q",
      "evidence": "Response time delay confirmed",
      "tool": "manual",
      "timestamp": "2024-01-15T10:30:00Z"
    }
  ]
}

Step 2: Deduplicate Findings

Deduplication Strategies:

Strategy	When to Use	Example
CVE-based	Same CVE across hosts	Log4j on 10 servers
Root cause	Different symptoms, same issue	Multiple XSS from one sink
Endpoint	Same vuln, different params	SQLi in ?id= and ?user=
Component	Shared vulnerable library	jQuery 2.x across subdomains

Deduplication Process:

1. Group by CVE ID (exact match)
2. Group by vulnerability type + root cause
3. Group by affected component/library
4. Identify parent/child relationships
5. Mark as:
   - UNIQUE: Distinct vulnerability
   - DUPLICATE: Same as another finding
   - RELATED: Connected but separate

Example Deduplication:

## Before Deduplication (15 findings)

- CVE-2024-1234 on server01
- CVE-2024-1234 on server02
- CVE-2024-1234 on server03
- SQLi in /api/users?id=
- SQLi in /api/orders?id=
- SQLi in /api/products?id=
- XSS in search field
- XSS in comment field
  ...

## After Deduplication (5 unique findings)

1. CVE-2024-1234 (3 affected hosts)
2. SQLi in ID parameters (3 endpoints, same sink)
3. XSS - Stored (2 injection points, different handlers)
   ...

Step 3: Enrich with CVSS Scores

CVSS 3.1 Severity Levels:

Score	Severity	Priority
9.0-10.0	Critical	Immediate
7.0-8.9	High	24-48 hours
4.0-6.9	Medium	1-2 weeks
0.1-3.9	Low	Scheduled
0.0	None	Informational

Calculating Custom CVSS:

For findings without CVE, calculate CVSS manually:

Base Metrics:

- Attack Vector: N(etwork)/A(djacent)/L(ocal)/P(hysical)
- Attack Complexity: L(ow)/H(igh)
- Privileges Required: N(one)/L(ow)/H(igh)
- User Interaction: N(one)/R(equired)
- Scope: U(nchanged)/C(hanged)
- Confidentiality: N(one)/L(ow)/H(igh)
- Integrity: N(one)/L(ow)/H(igh)
- Availability: N(one)/L(ow)/H(igh)

Common Patterns:

- Unauthenticated RCE: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H = 9.8
- Auth SQLi: CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H = 8.8
- Reflected XSS: CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:C/C:L/I:L/A:N = 6.1
- Info Disclosure: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:N/A:N = 5.3

Step 4: Add EPSS Exploit Probability

EPSS Overview:

Exploit Prediction Scoring System (EPSS) v4

- Probability score: 0% to 100%
- Predicts likelihood of exploitation in next 30 days
- Data-driven ML model updated daily
- Higher = more likely to be exploited

EPSS Thresholds:

- > 50%: Very high likelihood, prioritize immediately
- 10-50%: Significant likelihood, high priority
- 1-10%: Moderate likelihood, standard priority
- < 1%: Low likelihood, can be scheduled

Fetching EPSS Scores:

# Single CVE lookup
curl -s "https://api.first.org/data/v1/epss?cve=CVE-2024-12345"

# Multiple CVEs
curl -s "https://api.first.org/data/v1/epss?cve=CVE-2024-12345,CVE-2024-67890"

# Response format
{
  "data": [
    {
      "cve": "CVE-2024-12345",
      "epss": "0.87523",
      "percentile": "0.99421"
    }
  ]
}

EPSS Integration:

| CVE           | CVSS Base | EPSS Score | EPSS Percentile | Priority    |
| ------------- | --------- | ---------- | --------------- | ----------- |
| CVE-2024-1234 | 9.8       | 87.5%      | 99th            | CRITICAL    |
| CVE-2024-5678 | 7.5       | 2.1%       | 65th            | MEDIUM      |
| CVE-2024-9012 | 9.1       | 0.3%       | 32nd            | HIGH (CVSS) |

Step 5: Check CISA KEV Catalog

Known Exploited Vulnerabilities:

CISA KEV Catalog Importance:

- Vulnerabilities actively exploited in the wild
- Federal agencies must remediate by deadline
- Highest priority for any organization
- If in KEV = Critical priority regardless of CVSS

KEV Check Process:

1. Download current KEV catalog (JSON)
2. Match CVE IDs against findings
3. Note remediation deadline
4. Flag as CRITICAL priority

KEV Catalog Check:

# Download KEV catalog
curl -s "https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json" \
  -o kev.json

# Check if CVE is in KEV
jq '.vulnerabilities[] | select(.cveID=="CVE-2024-12345")' kev.json

KEV Match Format:

## CISA KEV Match Found

**CVE-2024-12345** is in CISA Known Exploited Vulnerabilities catalog

- Vendor: Example Corp
- Product: Example Server
- Vulnerability: Remote Code Execution
- Date Added: 2024-01-10
- Due Date: 2024-01-31
- Notes: Active exploitation observed

**ACTION REQUIRED**: Remediate immediately (federal deadline applies)

Step 6: Apply Prioritization Matrix

Combined Scoring Matrix:

Priority Calculation:

CRITICAL (Score 100):

- In CISA KEV catalog, OR
- CVSS >= 9.0 AND EPSS > 50%

HIGH (Score 75-99):

- CVSS >= 9.0 with any EPSS, OR
- CVSS 7.0-8.9 AND EPSS > 10%, OR
- Confirmed exploitation in the wild

MEDIUM (Score 50-74):

- CVSS 7.0-8.9 with EPSS < 10%, OR
- CVSS 4.0-6.9 with any exploitation evidence, OR
- High business impact asset

LOW (Score 25-49):

- CVSS 4.0-6.9 with no exploitation evidence, OR
- CVSS < 4.0 on important assets

INFORMATIONAL (Score 0-24):

- CVSS < 4.0 on non-critical assets, OR
- Best practice violations

Priority Score Formula:

Base Score = CVSS_Score \* 10

Multipliers:

- KEV listed: +50 (automatic CRITICAL)
- EPSS > 50%: +30
- EPSS 10-50%: +20
- EPSS 1-10%: +10
- High-value asset: +15
- Internet-facing: +10
- Contains sensitive data: +10

Final Priority = min(Base Score + Multipliers, 100)

Step 7: Generate Triage Report

Triage Output Template:

# Vulnerability Triage Report

**Date:** YYYY-MM-DD
**Scope:** [Target description]
**Total Findings:** X
**Unique Vulnerabilities:** Y
**Duplicates Removed:** Z

---

## Executive Summary

| Priority | Count | % of Total |
| -------- | ----- | ---------- |
| Critical | X     | X%         |
| High     | X     | X%         |
| Medium   | X     | X%         |
| Low      | X     | X%         |
| Info     | X     | X%         |

---

## Critical Findings (Immediate Action)

### CRIT-001: [Vulnerability Name]

| Attribute | Value                 |
| --------- | --------------------- |
| CVE       | CVE-YYYY-NNNNN        |
| CVSS      | 9.8 (Critical)        |
| EPSS      | 87.5%                 |
| KEV       | Yes (Due: YYYY-MM-DD) |
| Affected  | [endpoints]           |

**Evidence:** [Proof of vulnerability]

**Recommendation:** [Specific fix]

---

## High Priority Findings

[Similar format for High priority items]

---

## Deduplicated Findings Summary

| Finding        | Occurrences | Root Cause      |
| -------------- | ----------- | --------------- |
| CVE-2024-XXXXX | 5 servers   | Single patch    |
| XSS variants   | 3 endpoints | Output encoding |

---

## False Positives Excluded

| Finding   | Reason for Exclusion |
| --------- | -------------------- |
| [Finding] | [Justification]      |

---

## Remediation Priority Order

1. [CRIT-001] - CVE-2024-XXXXX - Patch immediately
2. [HIGH-001] - SQLi in API - Code fix required
3. [HIGH-002] - Outdated TLS - Config change
   ...

Examples

Example 1: Scan Results Triage

Scenario: Processing 50 findings from automated scanner

# Triage: Nuclei Scan Results

**Input:** 50 findings from nuclei scan
**After Deduplication:** 23 unique vulnerabilities

## Priority Breakdown

### Critical (2)

1. **CVE-2024-29847** - Ivanti EPM RCE
   - CVSS: 10.0 | EPSS: 94.2% | KEV: Yes
   - Affected: epm.internal.example.com
   - Action: Patch to 2024 SU1 immediately

2. **CVE-2023-46747** - F5 BIG-IP RCE
   - CVSS: 9.8 | EPSS: 89.1% | KEV: Yes
   - Affected: lb01.example.com
   - Action: Apply hotfix + block mgmt port

### High (5)

1. **SQL Injection** - /api/v2/search
   - CVSS: 8.6 | No CVE (custom code)
   - EPSS: N/A | 3 endpoints affected
   - Action: Parameterized queries needed

2. **CVE-2024-21762** - FortiOS SSL VPN
   - CVSS: 9.6 | EPSS: 45.3% | KEV: No
   - Affected: vpn.example.com
   - Action: Upgrade FortiOS to 7.4.3+

[Continue for remaining findings...]

### Deduplicated (27 findings → grouped)

| Original Finding         | Count | Grouped As               |
| ------------------------ | ----- | ------------------------ |
| jQuery CVE-2020-11022    | 12    | LOW-001 (12 subdomains)  |
| Missing security headers | 8     | INFO-001 (all endpoints) |
| TLS 1.0 enabled          | 7     | MED-003 (7 servers)      |

Example 2: Multi-Tool Correlation

Scenario: Correlating findings from Burp, Nuclei, and manual testing

# Cross-Tool Triage Correlation

## Finding Correlation Matrix

| Vulnerability  | Burp | Nuclei | Manual | Status             |
| -------------- | ---- | ------ | ------ | ------------------ |
| SQLi in search | ✓    | ✓      | ✓      | CONFIRMED (3/3)    |
| XSS in profile | ✓    | ✗      | ✓      | CONFIRMED (2/3)    |
| SSRF attempt   | ✓    | ✗      | ✗      | NEEDS VERIFICATION |
| Log4Shell      | ✗    | ✓      | ✗      | FALSE POSITIVE     |

## Correlation Notes

### SQLi in Search (HIGH CONFIDENCE)

- Burp: Time-based blind SQLi confirmed
- Nuclei: Error-based detection triggered
- Manual: UNION-based extraction successful
- **Verdict:** CONFIRMED - Exploit exists

### XSS in Profile (MEDIUM CONFIDENCE)

- Burp: Reflected XSS in name field
- Manual: Stored XSS confirmed in bio field
- Nuclei: Template didn't trigger
- **Verdict:** CONFIRMED - Multiple vectors

### SSRF Attempt (LOW CONFIDENCE)

- Burp: Suspicious redirect behavior
- Not confirmed by other tools
- **Verdict:** NEEDS RETEST

### Log4Shell Detection (FALSE POSITIVE)

- Nuclei: Detected Log4j usage
- Verification: Version 2.17.1 (patched)
- **Verdict:** FALSE POSITIVE - Excluded

Example 3: Business Context Triage

Scenario: Prioritizing based on asset criticality

# Business-Contextualized Triage

## Asset Classification

| Asset                | Type       | Data Classification | Business Impact |
| -------------------- | ---------- | ------------------- | --------------- |
| payments.example.com | Production | PCI DSS             | Critical        |
| blog.example.com     | Production | Public              | Low             |
| staging.example.com  | Non-Prod   | Test Data           | Medium          |

## Adjusted Priorities

### Original vs. Adjusted

| Finding   | CVSS | Original Priority | Asset    | Adjusted Priority        |
| --------- | ---- | ----------------- | -------- | ------------------------ |
| XSS       | 6.1  | Medium            | payments | **HIGH** (PCI)           |
| RCE       | 9.8  | Critical          | staging  | **HIGH** (non-prod)      |
| SQLi      | 8.6  | High              | blog     | **MEDIUM** (public only) |
| Info Disc | 5.3  | Medium            | payments | **HIGH** (PCI data)      |

### Priority Adjustment Rules Applied

1. **PCI DSS systems**: +2 severity levels for any data exposure
2. **Non-production**: -1 severity level (still report)
3. **Public-only data**: -1 severity level
4. **Internet-facing**: +1 severity level
5. **Admin access**: +1 severity level

## Final Remediation Order

1. XSS on payments.example.com (PCI compliance risk)
2. Info Disclosure on payments.example.com (PCI data)
3. RCE on staging.example.com (lateral movement risk)
4. SQLi on blog.example.com (public content only)

Error Handling

Issue	Cause	Resolution
No CVE available	Zero-day or custom code	Calculate manual CVSS score
EPSS not found	New CVE not in database	Use CVSS only, check again later
KEV check fails	API/connectivity issue	Use offline KEV JSON file
Duplicate CVE IDs	Same vuln, different endpoints	Group by CVE, list all affected
Conflicting severity	Tools report different CVSS	Use NVD authoritative score
Unknown asset context	No business classification	Default to HIGH, flag for review
Stale findings	Old scan results	Re-verify before reporting

Tool Reference

Resource	Purpose	URL
NVD	Authoritative CVSS	nvd.nist.gov
FIRST EPSS	Exploit probability	first.org/epss
CISA KEV	Exploited vulnerabilities	cisa.gov/known-exploited-vulnerabilities
CVSS Calculator	Manual scoring	first.org/cvss/calculator/3.1
CWE Database	Vulnerability types	cwe.mitre.org

Triage Decision Tree

START: New Finding
  │
  ├─ Has CVE? ─────────────────────────────┐
  │   │                                     │
  │   ├─ YES: Check CISA KEV               │
  │   │   │                                │
  │   │   ├─ In KEV? → CRITICAL            │
  │   │   │                                │
  │   │   └─ Not in KEV → Get EPSS         │
  │   │       │                            │
  │   │       ├─ EPSS > 50%? → Use CVSS+30 │
  │   │       ├─ EPSS 10-50%? → Use CVSS+20│
  │   │       └─ EPSS < 10%? → Use CVSS    │
  │   │                                     │
  │   └─ NO: Calculate Manual CVSS          │
  │       │                                 │
  │       └─ Apply business context         │
  │                                         │
  └─ Check for Duplicates ──────────────────┘
      │
      ├─ Exact duplicate? → Merge
      ├─ Same root cause? → Group
      └─ Unique? → Keep separate
          │
          └─ Apply Prioritization Matrix
              │
              └─ OUTPUT: Triaged Finding

References

• CVSS 3.1 Specification
• EPSS Model and API
• CISA KEV Catalog
• NVD Vulnerability Database
• CWE Database
• CVSS Calculator
• EPSS API Documentation
• Vulnerability Prioritization Guide