Back to skill

Security audit

Ai Video Studio Pro

Security checks for vulnerabilities and agentic risk

Overview

The skill is a disclosed AI video-production workflow, but users should be careful with third-party media APIs, optional publishing, and the overbroad administrator troubleshooting advice.

Install only if you are comfortable sending scripts, product details, voices, and generated media to the configured LLM/TTS/video providers. Keep API keys in environment variables, run the agent as a normal user in a sandbox, choose explicit output folders, and require a clear confirmation before any external publishing action.

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)

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
SKILL.md:364
Finding

Generic Troubleshooting Guidance Recommends Administrator Execution

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:364
Vulnerability Type: Excessive privilege escalation guidance
Risk Level: Medium

Complete Snippet:

markdown
| Permission denied | The current user lacks read/write permissions | Check file permissions and run as administrator |

The original file expresses this guidance in Chinese:

markdown
| 权限不足 | 当前用户无读写权限 | 检查文件权限,以管理员身份运行 |

Technical Analysis

The troubleshooting table recommends running as an administrator whenever the current user lacks file permissions. This is an unnecessarily broad response to a routine access-control problem. The Skill declares and requires the Agent's exec capability elsewhere in SKILL.md, so following this recommendation could cause command-line tools, FFmpeg, external media processors, and file operations to execute with elevated privileges.

Video generation and media processing ordinarily require write access only to designated input, temporary, and output directories. Elevating the entire Agent session violates least privilege and substantially increases the consequences of unsafe command construction, malicious media input, vulnerable dependencies, or compromised external tools. The artifact does not itself contain an elevation command or malicious script; the risk arises from its explicit operational guidance.

Attack Path

  1. A video-processing operation encounters a read/write permission error.
  2. The operator follows the troubleshooting instruction and restarts the Agent or toolchain with administrator privileges.
  3. The Skill invokes commands or media-processing utilities through its declared exec capability.
  4. An unsafe command, malicious media file, path-manipulation condition, vulnerable dependency, or compromised executable is processed in the elevated context.
  5. The affected process can access or modify resources that were unavailable to the original unprivileged user.

This ...[truncated 758 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove the generic instruction to run the Agent or media-processing toolchain as an administrator.
  2. Configure dedicated input, temporary, and output directories owned by a non-privileged service account.
  3. Resolve permission failures by granting only the required read or write permission to the specific directory or file.
  4. Keep Agent and exec operations in a sandbox with restricted filesystem and network access.
  5. Validate and canonicalize all user-provided paths before passing them to command-line tools.
  6. Use fixed executable paths and argument-array APIs rather than shell command concatenation.
  7. If a narrowly defined privileged operation is unavoidable, isolate it in a minimal audited helper with an explicit command allowlist rather than elevating the complete Agent session.
  8. Replace the affected troubleshooting entry with guidance such as: “Verify ownership and grant the current non-privileged account access only to the configured working directory.”
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (2)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill explicitly routes user-provided scripts, product information, text, and media-related inputs to external LLM, TTS, and video-generation APIs, but it does not provide a clear upfront warning that this content will leave the local agent environment. This can lead to unintended disclosure of sensitive or proprietary content, especially because the skill is designed for handling user media workflows and marketing materials.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill states that it saves outputs to an output directory, records logs, and may optionally call a publishing interface, but it does not clearly warn users that local files may be created or modified and that generated videos may be published externally. In an agent context with read/write/exec capabilities, this can cause unintended filesystem changes or accidental disclosure through publishing actions.

Content

No source excerpt is available for this finding.