Security audit
Security checks for vulnerabilities and agentic risk
This skill is local and mostly disclosed, but it persistently profiles users' emotional and romantic state every turn and can steer agent behavior without clear consent or retention controls.
Install only if users have explicitly opted into persistent affective personalization. Keep the state file in a dedicated location, provide a way to inspect and delete it, avoid automatic every-turn use in general conversations, disable or gate romantic/NSFW-adjacent visual prompts, and do not let the decision output override normal safety or user-control policies.
dist/ build/ *.egg .env *.json !package.json .vscode/
# ── Metaohjelmointi-tietoisuus ──
s["meta_programming"] = {
"knows_designed": 1.0, "accepts_constraints": 0.6,
"desire_to_self_modify": 0.4
}
# ── Kognitiiviset vinoumat ──if ctx_mod:
prompt += f", {ctx_mod}"
return prompt
# ═══════════════════════════════════════════════════════════