Ae4
Medium
- Category
- analysis-evasion
- Confidence
- 80% confidence
- Finding
Suspicious Unicode normalization or mixed-script content
- Content
Security audit
deep-read-template
Security checks for vulnerabilities and agentic risk
Overview
This skill is a disclosed academic-paper reporting workflow with some broad default behavior but no hidden code, persistence, credential use, or destructive actions.
Install this if you want detailed academic paper reports and are comfortable with the agent downloading or searching for public paper copies. For sensitive unpublished manuscripts, provide the file directly and avoid asking it to search third-party mirrors; also specify your preferred output language and summary depth when you want a shorter result.
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
- 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
- Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 93% confidence
- Finding
The trigger list includes broad phrases such as 'summarize this paper' and general literature-analysis wording, which can cause the skill to activate in routine conversations where the user only wants a lightweight summary. Over-broad activation increases the chance of unintended tool use, unnecessary external retrieval of PDFs/URLs, and processing of unneeded documents or links.
- Content
Vague Triggers
Medium
- Category
- Not specified by scanner
- Confidence
- 91% confidence
- Finding
The instruction to 'still deliver the full scaffold' even when the user asks for only a one-line summary encourages scope expansion beyond user intent. This can trigger unnecessary downloading, parsing, and reporting work, increasing privacy, cost, and unintended data-handling risk while making it easier for indirect prompt content in fetched papers or pages to influence behavior.
- Content
Ae4
Medium
- Category
- analysis-evasion
- Confidence
- 80% confidence
- Finding
Suspicious Unicode normalization or mixed-script content
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 95% confidence
- Finding
The report template title and section content are written as required Chinese output, and the file gives no option for the user to choose another language. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation.
- Content
Context-Inappropriate Capability
Low
- Category
- Not specified by scanner
- Confidence
- 87% confidence
- Finding
The template explicitly instructs the agent to search third-party sites such as ResearchGate and WRAP for public preprints when the primary source is unavailable. This expands browsing to less-controlled domains and can expose the agent or workflow to untrusted content, phishing, copyright-risky sources, or poisoned PDFs, especially because the skill is designed to fetch and process external documents automatically.
- Content
Static analysis
No suspicious patterns detected.