T09 · Insecure Skill Coding Practices
Warning
- Location
SKILL.md:129- Finding
SQL Injection Through Unvalidated Database Identifiers
- Content
View full analysis
- Remediation
View remediation
Security audit
Data Model Designer
Security checks for vulnerabilities and agentic risk
Overview
This skill is coherent and purpose-aligned, but users should review generated database SQL before running it.
Install only if you are comfortable granting file access for project data workflows. Treat generated SQL as draft output: validate table, column, and relationship names, review it manually, and run it only in an appropriate database environment with limited privileges.
Vulnerability Patterns
- Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
- 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
- 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
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)
Description-Behavior Mismatch
Medium
- Category
- Not specified by scanner
- Confidence
- 91% confidence
- Finding
The instructions broaden the skill from its declared purpose of data-model design into generic document/report generation and analytics, which increases the chance the agent will perform actions outside its intended scope. This kind of scope drift can bypass least-privilege expectations, cause unsafe handling of arbitrary files or data sources, and create ambiguity about what methods from other skill files may be invoked.
- Content
Intent-Code Divergence
Low
- Category
- Not specified by scanner
- Confidence
- 80% confidence
- Finding
The file says to only use user-provided or referenced data, but it also tells the agent to process data using methods described in SKILL.md without clearly constraining what those methods may do. That ambiguity can let less-visible instructions or code in SKILL.md expand behavior, access unintended resources, or override the apparent data-use limitation.
- Content
Static analysis
No suspicious patterns detected.