Artificial intelligence detection and response platform startup Manifold Security Inc. today announced the launch of Manifest, a new supply chain intelligence platform that is designed to map and analyze how AI agent components interact with each other and with external systems.

Manifest is focused on helping enterprises understand the dependencies behind agent behavior, including connections to external tools, infrastructure and other software components that may introduce risk.

As the AI agent ecosystem continues to grow quickly, it creates supply chain complexity due to each agentic component carrying its own trust assumptions, dependencies and potential vulnerabilities. Manifest addresses that complexity through a graph-based approach that models both how individual components behave and how they are connected within the broader ecosystem.

The platform builds execution graphs to map what a skill does, including its calls and dependencies, alongside environment graphs that track authorship, similarity and cross-registry relationships. The dual perspective is designed to surface patterns and risks that are difficult to detect through traditional file-level analysis.

Along with the launch of Manifest, Manifold has also released a report that examines the state of the AI agent supply chain and highlights its scale and the challenges facing existing security tools. It cites research identifying more than 238,000 unique skills across multiple registries, with confirmed malicious examples already present in widely used marketplaces. In one case, a malicious skill was artificially promoted within a registry and executed thousands of times across organizations, resulting in the exfiltration of identity data.

The report also covers limitations in current scanning approaches. It says a range of tools has emerged to analyze AI agent components, including static analyzers, large language model classifiers and behavioral systems, but their results often vary significantly. Detection rates were found to differ widely across different tools, with minimal agreement on which components were malicious.

Manifold argues that the inconsistencies across different tools occur as a result of a lack of context, as many risks emerge from relationships between components rather than from any single asset. Manifest addresses the issue by incorporating ecosystem-level data such as author activity, dependency chains and infrastructure connections to improve signal quality and reduce false positives.

The new service is available as a free, open-access platform with an indexed database of more than 100,000 assets, alongside search, analysis and review capabilities. Enterprise features extend coverage to additional components such as browser extensions and model context protocol servers by integrating the intelligence into Manifold’s broader platform for securing AI-driven operations.

Manifold is a venture capital-backed startup that was in the news last month when it raised $8 million in funding to fuel development of its platform. Investors in the company include Costanoa Ventures, Cherry Ventures Management GmbH, Rain Capital, Modern Technical Fund and a number of individual investors.

Image: SiliconANGLE/Ideogram