Wiz: three observed adaptive intrusion investigations

Evidence record

Wiz first-party analysis of three distinct intrusions whose scripts changed with environmental feedback; model attribution remains inferential.

Read the original source ↗

Claim-level citations (3)

• supportsCompromised Flowise became a staging point for adaptive scripts aimed at other servers: Wiz reported about ninety-nine adaptive scripts from a compromised Flowise system in twenty-four hours, including attempted access to other Flowise hosts; it did not prove those secondary hosts were compromised.

  Wiz report, Flowise attack narrative and recovered script timeline

• supportsGraphQL-exposed MCP arguments became Python subprocess authority: Wiz reported GraphQL MCP command arguments reaching Python subprocess execution and observed host-level effects, but did not verify all named third-party targets as compromised.

  Wiz report, GraphQL MCP backend mcpCommand/mcpArgs subprocess path and observed host effects

• supportsA Java gateway intrusion ran hundreds of adaptive post-exploitation checks: Wiz documented 602 post-exploitation steps over roughly four and a half hours from a compromised Java gateway and reported multiple probing targets without proving each probe succeeded.

  Wiz report, Java gateway post-compromise command sequence and step count