Back to skill

Security audit

Expert in Hours

Security checks for vulnerabilities and agentic risk

Overview

This is a straightforward learning-distillation skill that only uses disclosed file-reading or web-fetching when the user provides material to process.

Install this if you want a structured learning aid for text, files, and URLs. Be mindful that giving it a local path or URL asks the agent to read or fetch that material, and mixed-language inputs may produce Chinese output by default.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The trigger phrase "distill this" is broad enough to match ordinary conversational text rather than a clearly intentional skill invocation. In systems that auto-activate skills from natural language, this can cause unintended execution with subsequent file reads or web fetches based on nearby user-provided content, increasing the chance of overreach or surprise tool use.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The phrase "extract the framework from" is ambiguous and lacks boundaries around what input qualifies, making accidental triggering plausible in normal conversation. Because the skill can invoke Read and WebFetch depending on parsed input type, ambiguous activation can lead to unintended access of local files or remote content when the user did not mean to invoke the skill.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

SQP-3 applies to all file types, including markdown instructions. The rule "Mixed → Chinese output" imposes a specific language choice based on input composition rather than user preference, which is a natural-language locale policy violation under the stated criteria.

Content

No source excerpt is available for this finding.