Back to skill

Security audit

MiniMax H3 视频工具箱|AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

This appears to be a real AI-HIVE video workflow skill, but it needs review because it stores API keys locally and can send them to a configurable API endpoint without validating the destination.

Install only if you are comfortable with a Chinese AI-HIVE video workflow that reads local media you choose, uploads it to AI-HIVE, may create paid generation tasks, and stores an API key locally. Prefer using AI_HIVE_API_KEY for a session instead of init, do not use custom base URLs unless you trust them, and rotate the API key if it may have been used with an untrusted endpoint.

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:90
Finding

Bearer credential can be transmitted to an attacker-controlled API endpoint

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Note
Location
scripts/videogen.py:7
Finding

Runtime dependency installation is unpinned and lacks integrity verification

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (12)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The declared description presents a full-featured video production toolbox with unified mode selection and external API-driven generation operations. The supplied code does not implement those behaviors. It only parses command-line inputs and saves a static JSON blueprint skeleton. While the output fields loosely align with the domain and naming in the description, the actual implemented capability is materially narrower and lacks the key operational features described, so this is a clear description-behavior mismatch.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description presents a high-level generative video toolbox centered on MiniMax/Hailuo H3 and AI-HIVE OpenAPI interactions for creating and managing video generation jobs. The supplied code does none of that. Instead, it is a straightforward local CLI wrapper around ffmpeg/ffprobe for basic editing operations on existing video files. This is a materially different primary purpose and omits the declared API/networked generation behavior entirely. Therefore the description does not accurately represent the code chunk.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding

The skill advertises and demonstrates capabilities that imply network access, shell execution, file access, and environment-variable use, but it does not declare any explicit tool scope or permission boundaries. In an agent environment, this can lead to overbroad tool availability, making it easier for prompt-driven workflows to access local files, invoke shell commands, or use secrets such as API keys without clear least-privilege controls.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill enables implicit invocation without clear trigger constraints, which can cause the agent to activate this capability in response to loosely related user requests. Because the skill can produce runnable workflows and AI-HIVE commands tied to external content generation, unintended activation may lead to unnecessary API actions, cost-incurring operations, or execution of a workflow the user did not explicitly request.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script description, skill name, default platform value, and several embedded content fields are all fixed in Chinese, with no option for the user to choose language or locale. This creates a natural-language policy concern because the file imposes a specific language context rather than offering an explicit opt-in or choice.

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
94% confidence
Finding

The top-level docstring says this file is a '通用 AI 能力调用工具' that supports text chat, image generation, video generation, model queries, media upload, and task polling. However, the manifest for this skill claims a much narrower purpose centered on MiniMax H3 video-toolkit workflows; bundling broad chat/image/model/account functionality creates a semantic mismatch between the skill's claimed scope and the code it actually contains.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest focuses on receiving creative/video inputs and producing MiniMax H3 video generation suggestions, tasks, pricing snapshots, uploads, polling, and downloads. The code also implements a user-info endpoint and CLI command that retrieves account information and wallet balance, which is not justified by the described creator-toolkit workflow.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The manifest is narrowly framed around MiniMax H3 video workflows—text-to-video, image-to-video, and reference-based video generation guidance and execution. The code nonetheless includes a full text-chat API wrapper and corresponding CLI path, which is a broader AI capability not needed for the stated specialized video-toolkit purpose.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The manifest emphasizes MiniMax H3 video creation modes and related execution steps, not standalone image generation. This code exposes a general image generation endpoint and command, expanding the capability set beyond what is necessary for the declared video-only toolkit use case.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The init flow persistently stores the API key in ~/.ai-hive/config.json after interactive entry, but does not explicitly warn the user that credentials will be written to disk. Although the file is chmod'd to 0600, persistent local storage still increases credential exposure risk through backups, endpoint compromise, shared home-directory scenarios, or accidental inclusion in support bundles.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The skill parser description labels this as an 'AI Hive 裸接口视频 Skill', implying a video-focused interface. Yet the same skill exposes additional non-generation commands such as raw media upload and task inspection, which broadens the operational intent expressed by the description.

Content

No source excerpt is available for this finding.