Back to skill

Security audit

视频号爆款短视频拆解|AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

This skill mostly matches a video-generation workflow, but it handles API keys and media transfers through insufficiently constrained network endpoints.

Install only if you are comfortable giving this skill an AI-HIVE API key and letting it upload selected media and download generated results. Avoid custom base URLs unless you fully control them, review environment variables and ~/.ai-hive/config.json before use, prefer --no-download for dry runs, and confirm any paid generation parameters before execution.

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 (3)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/videogen.py:92
Finding

API Key Disclosure Through a Configurable Authentication Endpoint

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/videogen.py:278
Finding

Unvalidated Server-Controlled Media Upload and Download Destinations

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Note
Location
scripts/videogen.py:17
Finding

Unpinned Third-Party Dependency Installation Guidance

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • 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
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (16)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

A specialized analysis skill is presented, but the actual capability set reportedly includes generic chat, image/video generation, model querying, account/balance retrieval, and API-key initialization flows without any concrete logic for the promised breakdown outputs. This kind of scope deception can lead to privilege misuse, unexpected charges, and unsafe trust decisions because operators may treat it as a harmless analysis tool rather than a general-purpose API client.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

A specialized analysis skill is presented, but the actual capability set reportedly includes generic chat, image/video generation, model querying, account/balance retrieval, and API-key initialization flows without any concrete logic for the promised breakdown outputs. This kind of scope deception can lead to privilege misuse, unexpected charges, and unsafe trust decisions because operators may treat it as a harmless analysis tool rather than a general-purpose API client.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
89% confidence
Finding

The skill advertises and demonstrates capabilities that require shell, filesystem, environment-variable, and network access, but it does not declare an explicit tool scope or permission boundary. That creates an overbroad and ambiguous execution surface where an orchestrator may grant more access than users expect, increasing the risk of unintended file access, secret exposure, or external requests.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The trigger list includes a very generic phrase, "带货视频", which can appear in many ordinary e-commerce or marketing conversations unrelated to this specific decomposition skill. Because the skill activates on mere mention of the phrase, the boundary between relevant and irrelevant requests is not specific enough.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

"微信视频号" is a broad platform name rather than a task-specific trigger, so many benign conversations about the platform could match it. The file does not clearly constrain invocation to decomposition, scripting, or authorized reference-video analysis when this phrase appears alone.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The default prompt hard-codes Chinese output ('production-ready Chinese workflow') without reflecting user preference or asking for consent. This can cause unintended language coercion, reduce usability for non-Chinese-speaking users, and increase the chance that users approve or execute generated workflows they do not fully understand.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

Implicit invocation is enabled with no visible trigger constraints, so the skill may activate on broad or ambiguous requests related to video marketing or short-video workflows. In this skill's context, that can route users into a workflow that handles reference-video decomposition and runnable AI-HIVE commands even when authorization, platform-policy compliance, or user intent has not been clearly established.

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
97% confidence
Finding

The file presents itself as a narrow 视频号爆款短视频拆解 skill, but the implementation is a reusable general-purpose AI Hive client supporting text chat, image generation, video generation, model enumeration, media upload, and task polling. This capability expansion violates least-privilege expectations: a caller or wrapper that trusts the manifest could unknowingly grant broader AI and account operations than the declared purpose requires.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The module docstring and the entire CLI/user-facing guidance are written only in Chinese, including setup and operational instructions. For a general-purpose skill described as a '通用 AI 能力调用工具', this effectively imposes a language/locale constraint without documenting an opt-in choice or region-specific justification.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The generic text-chat endpoint allows arbitrary model interaction unrelated to the stated decomposition/remake workflow. In a skill ecosystem, this creates a scope mismatch where downstream agents may route untrusted prompts through a capability users did not authorize, enabling unintended data processing or policy bypass via an overly broad backend client.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The image-generation capability is broader than the declared short-video decomposition scope and enables unrelated content creation from the same trusted skill surface. This increases misuse potential, especially because the skill metadata emphasizes analysis/rewriting of authorized reference videos rather than general media synthesis.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The script description and embedded skill metadata are entirely Chinese and target a specific Chinese platform context, but the file does not indicate that this locale/language scope is optional or user-selected. Under the policy, language constraints should be offered as a choice or clearly justified as region-specific.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

This code creates parent directories and writes JSON to the path supplied by --output, but there is no confirmation prompt, warning comment, or docstring explaining that it will modify the filesystem. Although the operation is simple and expected for a generator script, it still performs a file write and directory creation without explicit disclosure in the file itself.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

Exposing user/account info and wallet balance is out of scope for a video-decomposition skill and unnecessarily broadens access to sensitive account metadata. While not code execution, it increases privacy and reconnaissance risk if the skill is invoked in environments where users do not expect account inspection.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The skill automatically downloads generated results to a default local directory after task completion, which can write files to disk without an explicit confirmation at the skill interface. In a hosted agent or shared workstation context, this can surprise users, consume storage, or persist sensitive/generated media locally when they expected a dry-run or metadata-only operation.

Content

No source excerpt is available for this finding.