Security audit
Security checks for vulnerabilities and agentic risk
The skill is a coherent discussion agent with local adaptive memory and OpenAI API use, but users should understand the memory and third-party data flow before installing.
Install only if you are comfortable with a Ukrainian-first debate persona, local persistence of aggregate interaction state in memory.json, and user prompts being sent to OpenAI using an environment-provided API key. Avoid entering sensitive personal or confidential information unless that external processing is acceptable.
The language section declares 'Ukrainian (primary)' and 'English (secondary)' as fixed operating languages, but does not state that the user can choose or override the language. This can violate the language/locale policy because it imposes a language preference without explicit user opt-in or a documented region-specific justification.
The skill hard-codes Ukrainian as the primary language and English as secondary, overriding normal user language preference and potentially causing the agent to respond in an unexpected language. This can degrade usability, interfere with safety-critical communication, and reduce the user's ability to understand warnings or verify outputs.
The module persistently writes interaction-derived state to ./memory.json without any disclosure, consent flow, retention control, or indication of what data is being stored. Even though the stored fields appear limited in this snippet, silent persistence of behavioral history can create privacy risk, expand the blast radius of compromise, and violate user expectations in agent environments.
The user-facing reply strings are hard-coded in Ukrainian for both empty-input and error cases. This imposes a specific language regardless of the user's language preference, and the file does not document any justified locale restriction or offer a language choice.
The manifest declares supported language/locales as uk and en, but does not state that the user can choose or opt into one. Under the policy, forcing or predefining language behavior without explicit user choice can be a natural-language locale policy violation.
The classifier detects only Ukrainian keywords, and the canned responses are also fixed in Ukrainian. This imposes a language/locale behavior without user opt-in or documentation of a justified region-specific constraint.
The code sends both the raw user message and potentially sensitive internally generated reflection text to a third-party API. If users are not informed and the reflection can contain prior-memory-derived context, this creates a privacy and data-handling risk through undisclosed transmission and possible over-sharing.
The code embeds a hard-coded Ukrainian-language question string without any indication that the user selected or opted into that language. The policy for this audit flags language or locale constraints when a skill forces a specific language without user choice or documented justification.
The inline comment is written in Ukrainian ('рефлексія кожні 3 питання'), which reflects a language-specific choice without any indication of user opt-in or documented locale justification. This can violate organizational language/locale policy when a skill implicitly assumes a specific language.
This JSON is a manifest-like file, so vague-scope checks apply. The values "never_disclose" and "internal_only" describe restrictions, but they do not clearly specify when disclosure or usage is permitted, which can leave invocation or handling boundaries ambiguous to downstream systems or maintainers.
Dependencies lack version pinning, allowing potential malicious package updates. Consider pinning versions.
"license": "ISC",
"type": "commonjs",
"dependencies": {
"dotenv": "^17.2.3",
"node-fetch": "^3.3.2",
"openai": "^6.17.0"
}
Dependencies lack version pinning, allowing potential malicious package updates. Consider pinning versions.
"type": "commonjs",
"dependencies": {
"dotenv": "^17.2.3",
"node-fetch": "^3.3.2",
"openai": "^6.17.0"
}
}
Dependencies lack version pinning, allowing potential malicious package updates. Consider pinning versions.
"dependencies": {
"dotenv": "^17.2.3",
"node-fetch": "^3.3.2",
"openai": "^6.17.0"
}
}
This code invokes the handler with a fixed non-English input string, which suggests the skill or its test flow may be tied to a specific language context. The file does not indicate user opt-in, language selection, or a documented region-specific reason for the locale constraint.
No suspicious patterns detected.