Security audit
Security checks for vulnerabilities and agentic risk
This skill is a local token-budget checker that reads user-supplied text or files and does not show hidden persistence, network use, destructive actions, or privilege escalation.
Installers should expect a simple local CLI that can read a user-specified file to estimate token usage. Consider narrowing activation terms if accidental invocation would be annoying, and treat its token counts as rough estimates rather than exact model tokenizer results.
The trigger terms are very broad, including generic words like 'context' and 'token', which can cause the skill to activate in ordinary conversations unrelated to budgeting. This increases the chance of unintended invocation, unnecessary context processing, and user confusion, though it does not by itself enable code execution or direct data exfiltration.
The description advertises bilingual/Chinese behavior as part of the skill's operation without indicating that the user can opt in or out. This can override user language expectations, cause unwanted output shaping, and in agent systems may lead to policy drift or unnecessary prompt expansion.
Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
| Gated | Minimal privilege, fail-closed exit codes, dangerous ops denied by default |
### Safety boundary
- Static analysis only; never auto-executes external commands.
- On risk detection → FAIL (fail-closed); human decides.
- Does not delete / modify / publish anything on the user's behalf.
The file presents the skill name and primary description in both English and Chinese, but does not indicate whether this bilingual requirement is optional, user-selectable, or tied to a justified region-specific constraint. Under the stated policy, forcing a specific language or locale without opt-in can be a natural-language policy issue.
No suspicious patterns detected.