session-start-hook

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

• Third-party content exposure detected (high risk: 0.80). The skill's startup hook can fetch and install external skill repositories via the skills_stale check and the command npx skills install "$SKILLS_REPO" (defaulting to camacho/ai-skills and overridable via AI_SKILLS_REPO), which would bring in untrusted third-party code that can materially change agent behavior.

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

• Potentially malicious external URL detected (high risk: 0.80). The hook can run npx skills install "$SKILLS_REPO" at runtime (defaulting to camacho/ai-skills or whatever AI_SKILLS_REPO is set to), which fetches and installs remote skill code that may execute and/or inject agent-controlling prompts, so this is a runtime external dependency that can control execution.