Ae1
High
- Category
- analysis-evasion
- Confidence
- 100% confidence
- Finding
Referenced artifact was not completely inspected
- Content
md node scripts/run.mjs --sample # 先看样例
Security audit
社保公积金核对（免费版）
Security checks for vulnerabilities and agentic risk
Overview
This is a local social-insurance checking skill with no evidence of hidden network access, persistence, or data movement.
Use this only for local checking of copied social-insurance table text, and do not treat it as legal, tax, audit, or policy advice. Be aware that the free version does not perform ratio, base-limit, or eligibility checks even though one summary phrase is imprecise.
Vulnerability Patterns
- Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
- Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
- Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
- Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
- Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
- Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
- MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (10)
Ae1
High
- Category
- analysis-evasion
- Confidence
- 100% confidence
- Finding
Referenced artifact was not completely inspected
- Content
md node scripts/run.mjs --sample # 先看样例
Context-Inappropriate Capability
Medium
- Category
- Not specified by scanner
- Confidence
- 87% confidence
- Finding
Granting Bash to a skill that is described as a local document-checking utility gives the skill broad command-execution capability beyond what users would reasonably expect from a read-oriented validation tool. If the runtime permits arbitrary shell commands, a malicious or compromised script could read unrelated local files, modify data, or invoke external programs despite the skill's benign business context.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 88% confidence
- Finding
The file’s natural-language documentation and all user-facing explanatory text are written entirely in Chinese, with no indication that another language is supported or that the user can choose their preferred language. Under the policy rule, forcing a specific language without user opt-in is a locale/language policy concern.
- Content
Intent-Code Divergence
Medium
- Category
- Not specified by scanner
- Confidence
- 82% confidence
- Finding
The header comment states this free version only implements free checks and that withheld checks are merely descriptive text, not implemented. However, the code later accepts
payload.rates,baseMin, andbaseMaxand reportsbasis: '...比例按缴费基数复算', which signals ratio-based checking capability despite those paid checks not actually running.- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The file’s natural-language interface is presented entirely in Chinese, beginning with the capability name and continuing through help text and runtime messages. There is no indication that the skill is region-specific by policy or that users may opt into another language, which can violate a language/locale choice requirement.
- Content
Description-Behavior Mismatch
Low
- Category
- Not specified by scanner
- Confidence
- 76% confidence
- Finding
The manifest repeatedly narrows behavior to a deterministic, local free subset with specific excluded checks (L05, L11, L46-L53), yet the declared tooling includes Bash at L16, which can execute arbitrary local commands beyond those stated checks. In this file there is no implementation or documented constraint showing Bash is limited to invoking only the packaged checker, so the declared behavior is broader than the described free-check-only scope.
- Content
Intent-Code Divergence
Low
- Category
- Not specified by scanner
- Confidence
- 76% confidence
- Finding
The module comment says the tool will not make determinations with insufficient materials and will not apply default values. Yet in normal execution it always merges user input with
DEFAULT_RATESat L269, meaning default ratio values are applied even though ratio checks are not part of the free checks actually executed.- Content
Description-Behavior Mismatch
Low
- Category
- Not specified by scanner
- Confidence
- 83% confidence
- Finding
The result summary states a basis of '比例按缴费基数复算' even though the free version explicitly does not execute ratio validation. In a compliance-checking skill, overstating what was actually checked can mislead users into relying on controls that never ran, causing false assurance in payroll or statutory reporting workflows.
- Content
Natural-Language Policy Violations
Low
- Category
- Not specified by scanner
- Confidence
- 91% confidence
- Finding
The JSON contains only Chinese headers and names in the sample text, which reflects a fixed language/locale choice. For a general-purpose template file, this can violate language/locale policy because it does not offer alternatives or explain that the skill is intentionally China-specific.
- Content