---

name: codeql description: Run CodeQL static analysis for security vulnerability detection, taint tracking, and data flow analysis. Use when asked to analyze code with CodeQL, create CodeQL databases, write custom QL queries, perform security audits, or set up CodeQL in CI/CD pipelines. allowed-tools:

• Bash
• Read
• Glob
• Grep

---

CodeQL Static Analysis

When to Use CodeQL

Ideal scenarios:

• Source code access with ability to build (for compiled languages)
• Open-source projects or GitHub Advanced Security license
• Need for interprocedural data flow and taint tracking
• Finding complex vulnerabilities requiring AST/CFG analysis
• Comprehensive security audits where analysis time is not critical

Consider Semgrep instead when:

• No build capability for compiled languages
• Licensing constraints
• Need fast, lightweight pattern matching
• Simple, single-file analysis is sufficient

Why Interprocedural Analysis Matters

Simple grep/pattern tools only see one function at a time. Real vulnerabilities often span multiple functions:

CopyHTTP Handler → Input Parser → Business Logic → Database Query
     ↓              ↓              ↓              ↓
   source      transforms       passes       sink (SQL)

CodeQL tracks data flow across all these steps. A tainted input in the handler can be traced through 5+ function calls to find where it reaches a dangerous sink.

Pattern-based tools miss this because they can't connect request.param in file A to db.execute(query) in file B.

When NOT to Use

Do NOT use this skill for:

• Projects that cannot be built (CodeQL requires successful compilation for compiled languages)
• Quick pattern searches (use Semgrep or grep for speed)
• Non-security code quality checks (use linters instead)
• Projects without source code access

Environment Check

Copy# Check if CodeQL is installed
command -v codeql >/dev/null 2>&1 && echo "CodeQL: installed" || echo "CodeQL: NOT installed (run install steps below)"

Installation

CodeQL CLI

Copy# macOS/Linux (Homebrew)
brew install --cask codeql

# Update
brew upgrade codeql

Manual: Download bundle from https://github.com/github/codeql-action/releases

Trail of Bits Queries (Optional)

Install public ToB security queries for additional coverage:

Copy# Download ToB query packs
codeql pack download trailofbits/cpp-queries trailofbits/go-queries

# Verify installation
codeql resolve qlpacks | grep trailofbits

Core Workflow

1. Create Database

Copycodeql database create codeql.db --language=<LANG> [--command='<BUILD>'] --source-root=.

2. Run Analysis

Copy# List available query packs
codeql resolve qlpacks

Run security queries:

Copy# SARIF output (recommended)
codeql database analyze codeql.db \
  --format=sarif-latest \
  --output=results.sarif \
  -- codeql/python-queries:codeql-suites/python-security-extended.qls

# CSV output
codeql database analyze codeql.db \
  --format=csv \
  --output=results.csv \
  -- codeql/javascript-queries

With Trail of Bits queries (if installed):

Copycodeql database analyze codeql.db \
  --format=sarif-latest \
  --output=results.sarif \
  -- trailofbits/go-queries

Writing Custom Queries

Query Structure

CodeQL uses SQL-like syntax: from Type x where P(x) select f(x)

Basic Template

Copy/**
 * @name Find SQL injection vulnerabilities
 * @description Identifies potential SQL injection from user input
 * @kind path-problem
 * @problem.severity error
 * @security-severity 9.0
 * @precision high
 * @id py/sql-injection
 * @tags security
 *       external/cwe/cwe-089
 */

import python
import semmle.python.dataflow.new.DataFlow
import semmle.python.dataflow.new.TaintTracking

module SqlInjectionConfig implements DataFlow::ConfigSig {
  predicate isSource(DataFlow::Node source) {
    // Define taint sources (user input)
    exists(source)
  }

  predicate isSink(DataFlow::Node sink) {
    // Define dangerous sinks (SQL execution)
    exists(sink)
  }
}

module SqlInjectionFlow = TaintTracking::Global<SqlInjectionConfig>;

from SqlInjectionFlow::PathNode source, SqlInjectionFlow::PathNode sink
where SqlInjectionFlow::flowPath(source, sink)
select sink.getNode(), source, sink, "SQL injection from $@.", source.getNode(), "user input"

Query Metadata

Key Language Features

Copy// Predicates
predicate isUserInput(DataFlow::Node node) {
  exists(Call c | c.getFunc().(Attribute).getName() = "get" and node.asExpr() = c)
}

// Transitive closure: + (one or more), * (zero or more)
node.getASuccessor+()

// Quantification
exists(Variable v | v.getName() = "password")
forall(Call c | c.getTarget().hasName("dangerous") | hasCheck(c))

Creating Query Packs

Copycodeql pack init myorg/security-queries

Structure:

Copymyorg-security-queries/
├── qlpack.yml
├── src/
│   └── SqlInjection.ql
└── test/
    └── SqlInjectionTest.expected

qlpack.yml:

Copyname: myorg/security-queries
version: 1.0.0
dependencies:
  codeql/python-all: "*"

CI/CD Integration (GitHub Actions)

Copyname: CodeQL Analysis

on:
  push:
    branches: [main]
  pull_request:
    branches: [main]
  schedule:
    - cron: '0 0 * * 1'  # Weekly

jobs:
  analyze:
    runs-on: ubuntu-latest
    permissions:
      actions: read
      contents: read
      security-events: write

    strategy:
      matrix:
        language: ['python', 'javascript']

    steps:
      - uses: actions/checkout@v4

      - name: Initialize CodeQL
        uses: github/codeql-action/init@v3
        with:
          languages: ${{ matrix.language }}
          queries: security-extended,security-and-quality
          # Add custom queries/packs:
          # queries: security-extended,./codeql/custom-queries
          # packs: trailofbits/python-queries

      - uses: github/codeql-action/autobuild@v3

      - uses: github/codeql-action/analyze@v3
        with:
          category: "/language:${{ matrix.language }}"

Testing Queries

Copycodeql test run test/

Test file format:

Copydef vulnerable():
    user_input = request.args.get("q")  # Source
    cursor.execute("SELECT * FROM users WHERE id = " + user_input)  # Alert: sql-injection

def safe():
    user_input = request.args.get("q")
    cursor.execute("SELECT * FROM users WHERE id = ?", (user_input,))  # OK

Troubleshooting

Rationalizations to Reject

Resources

• Docs: https://codeql.github.com/docs/
• Query Help: https://codeql.github.com/codeql-query-help/
• Security Lab: https://securitylab.github.com/
• Trail of Bits Queries: https://github.com/trailofbits/codeql-queries
• VSCode Extension: "CodeQL" for query development