advai-cli-unified-terminal-ai

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

• Third-party content exposure detected (high risk: 0.75). Skill installation/sync workflows accept GitHub URLs (e.g., advai skill install https://github.com/... / advai skill sync ...), which at runtime can fetch outsider-authored SKILL.md/skill text from public repos and ingest it into the agent’s LLM context.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly runs advai skill install against GitHub URLs (e.g., advai skill install https://github.com/your-org/skill-repo) which at runtime fetch repository "skill" definitions that become agent prompts/instructions, so remote content can directly control the agent.