Files
Anthropic-Cybersecurity-Skills/skills/detecting-sql-injection-via-waf-logs/references/api-reference.md
T
mukul975 27c6414ca5 Add folder anatomy (scripts/agent.py + references/api-reference.md) for 648 cybersecurity skills
Complete skill folder anatomy across all cybersecurity skills:
- scripts/agent.py: 80-150 line Python agents using real libraries (impacket,
  boto3, azure-mgmt-*, kubernetes, pefile, yara, scapy, shodan, stix2, etc.)
- references/api-reference.md: real API documentation with method signatures
- LICENSE: MIT license for all skill folders
2026-03-10 21:02:12 +01:00

1.9 KiB

API Reference: SQL Injection Detection via WAF Logs

ModSecurity Audit Log Sections

Section Content
A Audit log header (timestamp, transaction ID)
B Request headers (method, URI, HTTP version)
C Request body
E Response body
F Response headers
H Audit log trailer (rule matches, actions)

OWASP CRS SQLi Rules (942xxx)

Rule ID Description
942100 SQL Injection via libinjection
942110 SQL Injection (common keywords)
942120 SQL Injection operator detected
942130 SQL Injection tautology
942150 SQL Injection function detected
942160 Blind SQLi (sleep/benchmark)
942170 UNION query injection
942190 MSSQL code execution
942200 MySQL comment obfuscation
942210 Chained SQL injection
942280 PostgreSQL/MSSQL sleep
942290 MongoDB injection

SQL Injection Types

Type Pattern Severity
UNION-based UNION SELECT Critical
Time-based blind SLEEP(), BENCHMARK(), WAITFOR DELAY Critical
Error-based EXTRACTVALUE(), UPDATEXML() High
Tautology OR 1=1, AND 1=1 High
Stacked query '; DROP TABLE Critical
Schema enum INFORMATION_SCHEMA High
File access LOAD_FILE(), INTO OUTFILE Critical

AWS WAF Log Format (JSON)

{
  "httpRequest": {
    "clientIp": "203.0.113.42",
    "uri": "/api/users",
    "args": "id=1' OR 1=1--",
    "httpMethod": "GET"
  },
  "action": "BLOCK",
  "ruleGroupList": [{"ruleId": "SQLi_BODY"}]
}

Campaign Detection Logic

  • Group requests by source IP
  • Flag IPs with >= 5 SQLi attempts as campaigns
  • IPs with > 20 requests classified as automated tooling
  • Multiple attack types from same IP = multi-stage campaign

MITRE ATT&CK

  • T1190 - Exploit Public-Facing Application