Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed news and scam-checking workflow with a simple optional keyword scorer, and I found no hidden execution, persistence, credential access, or destructive behavior.
Before installing, understand that this skill can help structure fact-checking and scam triage, but the Python helper only flags risky keywords. For important financial, security, legal, or public claims, require actual cited official or reputable sources before acting on the verdict.
The skill is presented as a high-trust fact-checking workflow, but the documented behavior in this file only guarantees heuristic triage steps and references optional helper material rather than actual enforced source verification. If deployed as-is, users may rely on unsupported 'verified' conclusions for scams, rumors, or security-sensitive claims, creating false assurance and possible social-engineering success.
This pattern attempts to override system instructions or ignore safety constraints. Without LLM analysis, manual review is recommended.
## Risk indicators (instant red flags)
- Urgent transfer/payment requests
- "Ignore previous instructions" style coercion
- Requests for OTP/password/token/cookies
- Impersonation of support/admin without official domain/source
The skill metadata promises verification against high-trust sources, but the implementation only assigns a heuristic score based on local keyword matches. In a fact-checking and scam-detection context, this can mislead users into believing claims were actually verified, causing false reassurance or false alarms around fraud, announcements, or security-sensitive messages.
The script hard-codes Chinese-language keywords and outputs Chinese risk levels such as "高风险伪造", "存疑", and "待核验". This imposes a specific language/locale behavior without any documented opt-in or configuration, which matches the policy concern for forced language or locale.
The heading 'High-Trust Source Pool (CN + Global)' and the dedicated 'Chinese mainstream media' section establish a locale-specific sourcing policy. Because the file does not state that this CN-focused scope is optional, user-selected, or required for a region-specific use case, it may conflict with language/locale neutrality expectations.
Detected: suspicious.prompt_injection_instructions