Tp2
High
- Category
- MCP Tool Poisoning
- Confidence
- 85% confidence
- Finding
Mixing characters from multiple Unicode scripts in a single identifier is a common technique to create visually ambiguous tool names.
- Content
Security audit
Ai编程 0b1387
Security checks for vulnerabilities and agentic risk
Overview
This is a text-only skill with no executable payload, but its AI-programming purpose is polluted by unrelated merged content and triggers for buses, rail, electronics, and products.
Review before installing. It does not appear to run code or access your environment, but it is poorly scoped and contaminated with unrelated content, so it may activate in the wrong conversations or provide irrelevant guidance under an AI-programming label.
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
- 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 (7)
Description-Behavior Mismatch
High
- Category
- Not specified by scanner
- Confidence
- 98% confidence
- Finding
The skill claims to provide AI programming guidance, but large portions of the merged content are unrelated bus schedules, rail topics, electronics parts, and product listings. This semantic drift can cause the agent to activate in inappropriate contexts and return untrusted, irrelevant, or misleading content under a legitimate-looking skill name, undermining user trust and safety controls.
- Content
Vague Triggers
High
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The merged sections introduce unrelated trigger scopes such as rail and product terms, effectively expanding activation into domains completely unrelated to the declared skill. This creates a stronger security and safety problem than a normal broad trigger because a user can be routed into a mislabeled skill through arbitrary keywords, bypassing expectations about scope and provenance.
- Content
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The top-level trigger terms are broad phrases like 'AI编程' and related generic wording that are likely to appear in ordinary conversation. Overbroad triggers increase unintended activation risk, which is more dangerous here because the skill content is already contaminated with unrelated merged material and could surface misleading outputs in many benign chats.
- Content
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 96% confidence
- Finding
The activation rules include generic help-seeking language such as users needing 'AI programming related help,' which is ambiguous and likely to match many unrelated requests. In the presence of semantically drifted content, this broad scope can route users into a polluted skill and amplify misinformation or irrelevant guidance.
- Content
Intent-Code Divergence
Medium
- Category
- Not specified by scanner
- Confidence
- 94% confidence
- Finding
The document presents unrelated merged fragments as if they are versions or subskills of the same AI-programming capability, which misrepresents scope and provenance. That makes downstream routing, auditing, and user interpretation unreliable, increasing the chance that the agent serves irrelevant or unsafe content while appearing authoritative.
- Content
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 94% confidence
- Finding
The trigger phrase in this section is generic relative to the described domain and is attached to content that is itself unrelated to AI programming. This combination raises accidental invocation risk and can cause the system to answer with misleading content under the wrong skill identity.
- Content