rightstack

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

• Third-party content exposure detected (medium risk: 0.30). Runtime path run.mjs reads outsider-authored free text from process.env.SKILL_VAR (set via ${var} in SKILL.md) and passes it as the first argument to the pinned rightstack CLI (npx ... rightstack <command> --json).

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

• Potentially malicious external URL detected (high risk: 0.90). At runtime the skill runs npx to fetch and execute the pinned package rightstack@0.3.2 (spawned via npx --package rightstack@0.3.2 ...), so remote code is downloaded and executed and its output directly controls the agent's behavior.