Security audit
Security checks for vulnerabilities and agentic risk
This skill generates local conversation-rehearsal text and shows documentation inconsistencies, but no hidden access, persistence, network use, or destructive behavior.
Before installing, understand that the skill is a lightweight local text generator for rehearsing conversations. Its documentation is somewhat inconsistent about whether it is for real-world conversations or tabletop NPC writing, so users should verify it matches their intended use and avoid relying on it for legal, medical, violent, abusive, or other high-risk situations.
Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.
def _load_prompt_template(skill_name):
return PROMPT_TEMPLATE
def _clean_text(value: Any) -> str:
The manifest describes a skill for rehearsing real-world conversations with likely counterpart styles and explicitly frames it as help with boundaries, needs, and mildly stressful interactions. However, the documented scenarios on these lines repurpose the skill for fictional innkeeper NPC dialogue generation and NPC voice-diversity diagnosis, which is a different creative-writing/game-mastering function rather than real-world conversation support.
Without declared permissions the skill's intent is opaque and cannot be validated.
The top-level documentation says the skill is for practicing wording before a real conversation and respecting real relationships. That stated intent is contradicted by later examples showing innkeeper NPC scripting, villain improv, and session-note voice analysis, which are not real-world conversation rehearsal tasks.
The 'Use this skill when...' section includes broad everyday intents like asking for help, expressing a boundary, or preparing for a tense conversation, without explicit exclusions or negative examples. In a markdown skill description, this can make the trigger scope ambiguous and overlap with many common user requests unrelated to this specific rehearsal skill.
The manifest and overview consistently frame the skill as helping a user prepare for real conversations such as asking for help, setting boundaries, or handling mildly tense interactions. The later usage scenarios instead describe generating in-character dialogue for innkeepers and villains, plus analyzing RPG session notes, which is a materially different purpose than rehearsing real-world counterpart styles.
The title and usage scenarios switch between Chinese and English, and Scenario 1 is entirely presented in Chinese while other parts are in English. Because the README does not state that the skill adapts to the user's preferred language or lets the user choose, this can be read as imposing language behavior without explicit opt-in.
The file explicitly introduces a Chinese name and later includes a Chinese usage scenario and expected output, but does not state that language should follow the user's preference. This can create an implicit language/locale bias in skill behavior without opt-in.
The scenario and expected output are entirely in Chinese, but the skill description does not clarify whether output language should match user preference or input language. Under the language/locale policy rule, fixed-language guidance can be a policy concern when not explicitly optional.
No suspicious patterns detected.