Security audit

事实核查助手免费版

Security checks for vulnerabilities and agentic risk

Overview

The skill is a plausible fact-checking assistant, but it asks for broad shell, file, network, callback, and history capabilities without clear limits on what user content may be sent or stored.

Install only if you are comfortable with a fact-checking workflow that may send submitted claims or selected file contents to external search, fetch, fact-checking, LLM, or callback services. Avoid using it with private documents, credentials, regulated data, unpublished work, or confidential business material unless the skill is narrowed to scoped web tools, approved destinations, explicit consent, and clear history deletion controls.

A.I.G

Vulnerability Patterns

Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
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

Findings (2)

T09 · Insecure Skill Coding Practices

Warning

Location: SKILL.md:48
Finding: Uncontrolled Disclosure of User-Provided Content to External Services
Content: View full analysis
Remediation: View remediation

T05 · Unauthorized Access and Privilege Escalation

Note

Location: SKILL.md:19
Finding: Excessive Shell Execution and Local File Access Permissions
Content: View full analysis
Remediation: View remediation

SkillSpector

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (5)

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The activation language is so broad—covering general AI model calling, dialogue, agent orchestration, and LLM applications—that the skill may be invoked in many unrelated contexts. Overbroad routing increases the chance that sensitive user content is unnecessarily sent through a skill with shell-execution capability and ambiguous boundaries, expanding exposure and misuse risk.
Content: No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium

Category: Not specified by scanner
Confidence: 93% confidence
Finding: The skill claims to perform internet-based fact checking against external organizations, but its declared toolset only exposes local file/shell capabilities (read, exec, glob, grep) and no explicit web/network tools. This creates a dangerous mismatch: an agent may try to satisfy the documented behavior by using unrestricted shell execution to invoke network-capable binaries like curl, turning a benign research skill into a generic command-execution surface with network reach.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 92% confidence
Finding: The skill encourages saving and archiving user-submitted claims and content history without warning that submissions may contain personal, confidential, or regulated information. In a fact-checking context, users may paste private messages, drafts, or sensitive allegations, so silent retention materially increases privacy and data-handling risk.
Content: No source excerpt is available for this finding.

Intent-Code Divergence

Low

Category: Not specified by scanner
Confidence: 78% confidence
Finding: The line states that the capability supports '创建/查询/导出操作' (create/query/export), but the documented command set around it is about checking claims and viewing results. This is not just incomplete wording: 'create' suggests a broader record-management capability not reflected in the described fact-checking behavior.
Content: No source excerpt is available for this finding.

Intent-Code Divergence

Low

Category: Not specified by scanner
Confidence: 86% confidence
Finding: The manifest and overview consistently frame this as '免费版' and '面向个人用户提供基础的事实核查能力'. Line L135 expands coverage to '企业团队和自动化工作流场景', which directly conflicts with that stated scope and edition positioning.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.