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
Agent开发 09770e
Security checks for vulnerabilities and agentic risk
Overview
The skill does not contain code or destructive behavior, but its broad activation and unrelated merged content make it likely to trigger in the wrong context.
Install only if you are comfortable with a low-capability but messy learning note that may activate on broad Agent-related prompts. It should be narrowed to specific Agent-development tutorial triggers and split from unrelated PS Portal or game-review content before normal use.
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 (5)
Vague Triggers
High
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The trigger includes extremely broad phrases such as 'Agent开发' and especially 'Agent', which are likely to match ordinary conversation and unintentionally activate this skill. Because the skill content is already inconsistent, broad triggering makes the confusion more dangerous by increasing the chance of accidental invocation in unrelated contexts.
- Content
Vague Triggers
High
- Category
- Not specified by scanner
- Confidence
- 96% confidence
- Finding
The activation scenarios are vague ('用户说"Agent"' and general requests for Agent development help), which does not sufficiently constrain when the skill should run. In combination with duplicated and contradictory content, ambiguous activation increases the risk of the wrong instructions being surfaced or trusted during normal user conversations.
- Content
Description-Behavior Mismatch
High
- Category
- Not specified by scanner
- Confidence
- 97% confidence
- Finding
The manifest presents this as an Agent development learning skill, but later content shifts into unrelated gaming/device review material, creating a mismatch between declared purpose and actual instructions. In agent ecosystems, this is dangerous because routing and trust decisions are often based on manifest metadata; misleading metadata can trigger the wrong capability in response to common user requests.
- Content
Intent-Code Divergence
High
- Category
- Not specified by scanner
- Confidence
- 98% confidence
- Finding
The file redefines the same skill as unrelated PS Portal/SonyAgent content inside the documentation, which creates instruction ambiguity and weakens trust boundaries for downstream agents selecting or executing skills. This kind of content drift can cause the wrong skill to activate or supply irrelevant behavior under a trusted Agent-development label, increasing the risk of prompt confusion, misrouting, or unintended actions.
- Content
Static analysis
No suspicious patterns detected.