Back to skill

Security audit

无脸带货视频生成器|AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

The skill is not clearly malicious, but it needs review because it can store an AI-HIVE API key and send that credential to a configurable server while uploading selected media.

Review this before installing if you will use a real AI-HIVE key. Use only the official AI-HIVE base URL, check AI_HIVE_BASE_URL and ~/.ai-hive/config.json before running, prefer a limited or low-balance API key, and upload only media you are allowed to share. Install dependencies in a dedicated environment and be aware the workflow is primarily Chinese-language.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (2)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/videogen.py:91
Finding

API Bearer Credential Can Be Transmitted to an Arbitrary Configurable Endpoint

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Note
Location
scripts/videogen.py:17
Finding

Runtime Dependency Is Installed Without Version or Integrity Pinning

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (11)

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The declared description presents a full faceless shoppable-video generator workflow, including media generation and external API orchestration through AI-HIVE OpenAPI. The supplied code instead implements a local command-line utility that writes a draft JSON brief with project metadata. While the JSON includes labels and aliases consistent with the declared domain, the primary behavior is materially narrower and different: blueprint creation only. There are no external network calls, no media processing, no asynchronous job handling, and no output resembling generated video deliverables. Therefore the description substantially overstates the implemented functionality.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description promises an end-to-end faceless product-video generator with AI-assisted creation and API-based orchestration. The supplied code does not implement any of that. It only wraps ffmpeg/ffprobe to perform basic deterministic edit operations on existing local media files. While such editing could be a supporting component of a video workflow, this chunk’s primary behavior is materially narrower and different from the declared purpose, and the specifically claimed external API capabilities are entirely absent. Therefore this is a clear description-behavior mismatch.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding

The skill advertises and includes commands that use environment variables, local file access, network access, and shell execution, but it declares no explicit tool scope or permission boundaries. In an agent environment, that increases the chance the skill is invoked with broader capabilities than intended, enabling unintended filesystem access, outbound requests, or execution of user-influenced commands.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The default prompt hardcodes 'production-ready Chinese workflow' regardless of the user's language preference, which can override user intent and cause the agent to respond in an unexpected language. This is primarily a prompt-quality and UX control issue, but it can also reduce clarity for safety-critical instructions if the user cannot adequately review or validate the generated output.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The script description and multiple user-facing string literals are fixed in Chinese, with no option to select language or locale. This can violate language/locale policy when users are not given a choice or informed that the tool is limited to a specific language context.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/edit_video.py (reported line 20)May include surrounding context.

python
def run(args: list[str]) -> None:
    print(" ".join(args))
    subprocess.run(args, check=True)


def main() -> None:

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The module docstring and the code expose a general-purpose AI Hive toolchain including text chat, image generation, model listing, user-info lookup, arbitrary media upload, and generic task polling, not just faceless commerce video generation. While some of these are supporting primitives, the file clearly contains reusable capabilities unrelated to the manifest’s narrower purpose of generating faceless commerce/demo videos.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The module docstring is entirely in Chinese and presents the skill as a general-purpose tool, while later user-facing prompts and help text are also Chinese-only. For a general skill, forcing a specific language without opt-in or justification is a natural-language policy violation under the locale-policy rule.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The client implements get_user_info, and the generic CLI exposes user-info for querying user information and wallet balance. Accessing account metadata and balances is not justified by the manifest’s purpose of generating faceless commerce/product-demo videos for marketing content.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

This code creates parent directories and writes a JSON file to a user-specified path, but provides no prior confirmation, warning comment, or explanatory docstring about modifying the filesystem. Although the write is central to the script's purpose, the file path is user-controlled and the code gives no explicit disclosure before performing the write.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
83% confidence
Finding

The skill-specific CLI advertises and exposes task, upload, and init commands in addition to generate. The manifest focuses on accepting product inputs and delivering generated faceless commerce videos via the AI-HIVE workflow; standalone upload and generic task-inspection capabilities expand behavior beyond that user-facing description.

Content

No source excerpt is available for this finding.