using-vit

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The SKILL.md/COMMANDS.md explicitly tells the agent to run "vit skim --json" to read caps (public, user-generated ATProto records) from followed accounts and to consume those caps via commands like "vit remix" and "vit learn", so untrusted third-party content from the open ATProto network can directly influence the agent's actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill permits installing and loading remote skills/code at runtime (via commands like vit learn <ref> and probing/cloning beacons such as the example git URL https://github.com/org/repo.git), which would fetch external content that can install or execute code and thereby directly influence agent behavior.