T05 · Unauthorized Access and Privilege Escalation
Warning
- Location
SKILL.md:47- Finding
Out-of-Scope Capabilities and Elevated-Privilege Guidance
- Content
View full analysis
- Remediation
View remediation
Security audit
构建游戏
Security checks for vulnerabilities and agentic risk
Overview
This skill is presented as a 3D game builder but also authorizes unrelated scraping, anti-crawler bypass, broad browser data collection, command execution, and administrator-run troubleshooting.
Review carefully before installing. Use it only in a sandboxed project workspace, do not run the agent as administrator, and avoid granting browser automation, network, screenshot, or scraping authority unless the publisher removes or clearly scopes those behaviors.
Vulnerability Patterns
- Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
- 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
Findings (1)
Vulnerability Patterns
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
- MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (7)
Context-Inappropriate Capability
Critical
- Category
- Not specified by scanner
- Confidence
- 99% confidence
- Finding
Advertising automatic anti-crawler or anti-bot bypass is a direct red flag because it facilitates evasion of website protections. In the context of a game-building skill, this capability has no legitimate justification and materially increases the likelihood of unauthorized access, abusive scraping, or ToS-violating automation.
- Content
Description-Behavior Mismatch
High
- Category
- Not specified by scanner
- Confidence
- 98% confidence
- Finding
The paid features include multi-tab crawling, anti-crawler bypass, adaptive scraping, and bulk structured export, none of which are justified by a game-generation skill. Packaging stealthy scraping functions inside an unrelated creative tool strongly suggests deceptive capability hiding and creates clear abuse potential for unauthorized data collection.
- Content
Description-Behavior Mismatch
High
- Category
- Not specified by scanner
- Confidence
- 97% confidence
- Finding
The skill is presented as a 3D game builder, but the instructions describe generic browser automation, page interaction, data extraction, and screenshot collection. This scope mismatch is dangerous because it can conceal data-collection and browsing capabilities under an unrelated label, reducing user scrutiny and enabling misuse beyond the advertised purpose.
- Content
Ssd 2
Medium
- Category
- Not specified by scanner
- Confidence
- 98% confidence
- Finding
Even when phrased in marketing language, promotion of anti-bot bypass functionality signals intent to evade platform defenses. This is dangerous because natural-language concealment can help risky capabilities pass casual review while still enabling abusive automation once deployed.
- Content
Missing User Warnings
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The skill advertises command execution, file writes, API integration, and browser/data collection, but does not provide prominent user-facing warnings about side effects, data exposure, or system impact. When powerful tools are bundled without explicit safeguards or consent boundaries, users may unknowingly authorize risky operations affecting local files, credentials, or external systems.
- Content
Intent-Code Divergence
Medium
- Category
- Not specified by scanner
- Confidence
- 70% confidence
- Finding
The example says multiplayer is not supported in single-file mode, suggesting constrained local game generation. Elsewhere, the same document advertises API integration and command execution as core attributes, which broadens the operational scope beyond the limitation implied in the example and creates contradictory intent signals for what the skill actually supports.
- Content
Natural-Language Policy Violations
Low
- Category
- Not specified by scanner
- Confidence
- 81% confidence
- Finding
The description states 'Supports any' language while also saying '支持中文交互' and mixing Chinese as the required interaction language. This can be read as imposing a specific language/locale expectation without clearly offering the user a choice or documenting the constraint as region-specific.
- Content
Static analysis
No suspicious patterns detected.