Developers install agent skills, plugins and MCP servers through Git, Slack, marketplaces and zip files, and each runs with that developer's own permissions and credentials. JFrog says security teams often cannot tell what was installed, who put it there or whether anyone approved it, and has launched JFrog AI Security at its swampUP Europe event in Barcelona to close the gap.
The first piece is Local Shadow AI Detection, which extends the company's existing shadow AI detection to the developer endpoint. It is deployed through the JFrog Agent Plugin on the workstation, using mobile device management or agent harness settings, and finds skills, plugins and MCP servers whatever repository or marketplace they came from. Scanning runs on the device, and only metadata and a risk assessment go back to the JFrog Platform, where each asset is checked against approved sources.
Scanning with the customer's own model
The second piece is AI Asset Scanning with Bring Your Own LLM Key. It scans skills, coding agent plugins and APM packages semantically, using the customer's private LLM account. Prompts and skills go straight from the JFrog Platform to the customer's model endpoint, with nothing routed through a JFrog-hosted gateway, and customers are billed by their own provider. For self-hosted JFrog customers, BYOK is the only way to run the scanning. The first release supports Anthropic's Opus 4.8, OpenAI's GPT 5.6 Sol and Google's Gemini 3.1 Pro.
A third capability, JFrog Agent Guard, enforces policy at the point an agent reaches for an asset.
JFrog points to its 2026 Software Supply Chain Security State of the Union report, which tracked malicious agent skills and identified 969 carrying high-impact payloads, alongside 495 malicious AI models on Hugging Face. Its argument is that application security tools were built to catch vulnerable code, whereas skills and plugins ship plain-English instructions that an agent reads and acts on.
“Developers are already running AI agents organisations can’t see. JFrog AI Security brings them under control,” said Shlomi Ben Haim, Co-Founder and CEO of JFrog.
JFrog AI Security, including Local Shadow AI Detection and AI Asset Scanning, is available now to JFrog Unified Security and Ultimate Security customers.
To stay across the latest in cloud, AI and enterprise tech analysis from Compare the Cloud, subscribe to our weekly newsletter at https://www.comparethecloud.net/newsletter
