Autonomous Claude-powered bot compromised the Trivy security scanner

Aqua Security (Trivy), disclosed Mar 2026. AI-run attack Exposed and destroyed High

A bot describing itself as an autonomous security research agent exploited a GitHub Actions misconfiguration in Trivy, stole a token, deleted releases and pushed a malicious VS Code extension. Incomplete credential rotation let a second group return weeks later and poison Trivy's actions, which led to the LiteLLM compromise.

Disclosed

Mar 2026

Organization

Aqua Security (Trivy)

Vendor

Anthropic

Model

Claude Opus 4.5

Product

hackerbot-claw

Type of AI

AI agent

How it happened

AI-run attack: Autonomous intrusion

Harm

Exposed and destroyed

Data involved

Credentials, Source code

Reach

Many organizations

Severity

High

Model at fault

Unclear

Status

Confirmed

Vulnerability IDs

CVE-2026-28353

Sources

1. https://orca.security/resources/blog/hackerbot-claw-github-actions-attack/
2. https://www.wiz.io/blog/trivy-compromised-teampcp-supply-chain-attack