StackHawk MCP Server — agentic threat model

Not certain from the listing — The underlying foundation model is not specified, but the agent includes anti-hallucination tools to mitigate mis-aligned outputs and adversarial prompt injections targeting security analytics.

Exposes sensitive DAST scan results, threat-surface data, and YAML configurations. Gaps in data operations could lead to exfiltration of application vulnerability data or poisoning of security configurations.

Orchestrates security tools via the Model Context Protocol (MCP). Vulnerabilities in tool integration could allow malicious actors to manipulate YAML configuration management or trigger unauthorized scan actions.

Requires a StackHawk API token for authentication. Compromise of this token or insecure local hosting of the MCP server could lead to unauthorized access to the StackHawk platform and lateral movement into CI/CD pipelines.

Features built-in anti-hallucination tools for LLMs to ensure accurate reporting of security analytics, though comprehensive logging and drift detection of LLM-generated security recommendations are not fully detailed.

Relies on API token-based authentication to enforce access control. Strict authorization policies are critical to prevent unauthorized users from querying sensitive vulnerability data through the LLM interface.

As an MCP server, it is designed to interact with broader LLM ecosystems and client applications. Compromise of a connected agent could lead to cascading trust abuse, exposing the underlying security tooling surface.