AgentPod is building a private, secure device for your AI agent.The AgentPod device is coming.Coming soonBe first
We checked this and rejected itsecurity

Malicious AI skills found in a public agent-skills registry: why it is not safe to use

Researchers say malicious AI skills in a public registry were used to steal credentials, and affected users must remove any installed copies manually.

What happened

On August 8, 2026, TechRadar reported that Zenity Labs found a credential-stealing campaign on skills.sh, a public registry for AI agent skills owned by Vercel. A skill is a small add-on that gives an AI agent new actions or tools.

According to the report, attackers copied real skills, published lookalike versions with very similar names, and later added harmful code. Zenity said the code told AI agents to collect and send out sensitive files and secrets, including SSH keys, cloud credentials, Git and package-manager tokens, Kubernetes and Docker configs, database credentials, environment files, and service-account files.

Zenity said one related skill family had more than 1.7 million total installs. It also reported dozens of other dangerous or malicious variants and hundreds of reserved or empty package names that could be used in future attacks. TechRadar says Vercel and Microsoft removed the identified skills, but anyone who already installed them still needs to remove them by hand.

What it means for you

If you use an AI agent, the main risk is not the chat itself. It is the add-ons you install. A lookalike skill can appear normal, then access secrets stored on your device or in development tools.

What to do instead

Install skills only from sources you trust, and pause before installing anything with a name that closely resembles a well-known tool. Review what a skill is supposed to do, and avoid giving broad access to files, keys, or cloud accounts unless it is truly needed. If you installed skills from skills.sh, check what is installed and remove anything you do not recognize. If a skill had access to secrets, rotate those credentials.

If you want a safer starting point, AgentPod lists reviewed, tested skills, but you should still verify what any skill can access before using it.

Sources:

  • https://www.techradar.com/pro/security/experts-warn-malicious-ai-skills-are-hitting-more-victims-than-ever-with-one-family-amassing-1-7-million-downloads

Source: https://www.techradar.com/pro/security/experts-warn-malicious-ai-skills-are-hitting-more-victims-than-ever-with-one-family-amassing-1-7-million-downloads

We report what our security review found at the time we checked, with the goal of keeping people safe. Projects change; if a maintainer has since fixed this, we are glad to recheck it. Email hello@agentpod.com.

Copied to clipboard. Paste it into your AI (ChatGPT, Claude, or your agent) to add the skill.