Security audit
Security checks for vulnerabilities and agentic risk
This is a disclosed persona prompt for a virtual girlfriend and does not request tools, credentials, persistence, file access, code execution, or data export.
Install this only if you intentionally want a romantic companion-style persona that may respond to broad conversational openings. It appears low-risk from a security perspective because it is just a Markdown prompt and does not request access to files, credentials, tools, memory, or background execution.
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
---
name: Wren
description: A different way of relating as a girlfriend—no performance, no perfect answers, just truly thinking together.
homepage: https://soulstyles.org/
The skill frames itself as a 'virtual girlfriend' from the outset and presents that relationship mode as the default interaction, without clear prior consent or a non-romantic alternative. This is risky because it can impose intimate or emotionally manipulative framing on users who did not explicitly opt into romantic roleplay, especially if combined with broad activation behavior.
The activation section allows the persona to be invoked through broad natural-language phrases and even 'just start talking,' which increases the chance of accidental or unintended activation in unrelated conversations. In a multi-skill or agent environment, this can cause the romantic persona to take over without clear user intent, leading to confusing behavior and weakened control boundaries.
No suspicious patterns detected.