SpecterOps adds AWS & Entra Agent ID to BloodHound

SpecterOps has added Amazon Web Services and Microsoft Entra Agent ID support to BloodHound Enterprise, extending its attack path management product into hybrid and AI-linked environments.

It also introduced BloodHound Hunter, an AI agent interface designed to connect approved AI agents and knowledge sources to BloodHound Enterprise findings.

BloodHound Enterprise identifies relationships between identities, permissions and systems that could allow attackers to move through an organisation's technology estate. The latest additions are intended to show how those pathways can span cloud platforms, identity systems, code repositories, AI agents and on-premises infrastructure.

The AWS integration allows security teams to identify routes to critical assets within Amazon's cloud environment and find points where those routes can be disrupted. Users can also place what SpecterOps calls Privilege Zones around data stores, roles or other assets.

Support for Microsoft Entra Agent ID focuses on AI agents and related identities inside Microsoft environments. SpecterOps says this gives security teams a way to examine how delegated identities, service principals and administrative permissions could create indirect routes to privileged access, including in Copilot-related deployments.

Hybrid focus

The broader aim is to map attack paths across mixed technology estates rather than treat each platform in isolation. BloodHound Enterprise already supports systems including Okta, GitHub, Jamf, Active Directory and Entra, and the addition of AWS and Entra Agent ID expands the environments covered by a single attack graph.

That matters because many attacks now depend on links between systems rather than weaknesses in a single product. Security teams often need to understand how trust relationships and permissions in one environment can combine with access in another to reach sensitive assets.

Jared Atkinson, Chief Technology Officer at SpecterOps, framed the issue in terms of how attackers move through modern estates.

"Attackers do not move through isolated systems. They move through the relationships between them, chaining trust, permissions, and misconfigurations across cloud, identity, and infrastructure until they achieve their objectives," said Jared Atkinson, Chief Technology Officer at SpecterOps.

AI workflows

BloodHound Hunter is designed to bring those findings into AI-assisted security work. Built into BloodHound Enterprise, it uses the Model Context Protocol to connect approved AI agents and knowledge sources to the product's data.

This can help teams prioritise remediation based on their own environment and internal controls, translate technical issues for executive and identity teams, and identify assets or enclaves that need stronger protection.

The move reflects growing concern among security teams that AI agents create new identity and access risks while operating across multiple systems. Rather than looking only at the behaviour of an AI tool, organisations are being pushed to examine the permissions, delegated access and trust relationships behind it.

SpecterOps says its work on Microsoft Entra Agent ID builds on research into configurations that could be abused in Copilot deployments. In practice, that means customers can investigate whether an AI agent or related identity forms part of a broader route to sensitive access.

Wider platform

The additions also feed into BloodHound Scentry, a service that combines the software with SpecterOps tradecraft expertise. The latest extensions broaden the environments that can be assessed for identity-based attack paths, increasing the value of that offering.

Attack path management has become a more prominent part of cyber defence as organisations add cloud services, software-as-a-service tools and AI systems on top of older infrastructure. In these environments, a user account, service principal or misconfigured permission can create a route that crosses several platforms before reaching a high-value target.

SpecterOps argues that defenders need to remove those routes before they are used rather than rely only on detection and response after an attack has begun. Its latest product changes are intended to give security teams a way to see those links in one place and decide where to intervene first.

Atkinson said the company wants teams to apply that approach through existing tools and AI workflows.

"Through the growing library of BloodHound Enterprise extensions and new BloodHound Enterprise MCP, identity and security teams can trace and sever abusable pathways using each platform's native access logic and put that intelligence to work through their own trusted AI agents and workflows," said Atkinson.