Back to skill

Security audit

爆款视频多平台改版|AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a disclosed AI-HIVE video-generation helper, but it needs review because it can use a stored API key with an unvalidated custom API endpoint.

Install only if you understand the AI-HIVE API-key flow. Keep the API base URL fixed to the official AI-HIVE endpoint, do not follow prompts or examples that set --base-url or AI_HIVE_BASE_URL to another host, use a limited/rotatable API key, and confirm parameters and costs before running generation commands.

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)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/videogen.py:91
Finding

Bearer API Credential Disclosure Through an Unrestricted Base URL

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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 (18)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

If the skill exposes undeclared capabilities such as generic chat, image generation, account or wallet retrieval, or model catalog access beyond its stated purpose, users and reviewers cannot accurately assess its data flows or risk surface. Hidden or undocumented capabilities are dangerous because they can process unexpected data, contact external systems, or expose account information outside the user's informed expectations.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

If the skill exposes undeclared capabilities such as generic chat, image generation, account or wallet retrieval, or model catalog access beyond its stated purpose, users and reviewers cannot accurately assess its data flows or risk surface. Hidden or undocumented capabilities are dangerous because they can process unexpected data, contact external systems, or expose account information outside the user's informed expectations.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding

If the skill exposes undeclared capabilities such as generic chat, image generation, account or wallet retrieval, or model catalog access beyond its stated purpose, users and reviewers cannot accurately assess its data flows or risk surface. Hidden or undocumented capabilities are dangerous because they can process unexpected data, contact external systems, or expose account information outside the user's informed expectations.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding

The skill includes executable code paths and instructions that imply access to environment variables, local files, network requests, and shell utilities, but it does not declare any explicit tool scope or permissions boundary. This creates an authorization and review gap: a caller or platform may permit broader capabilities than intended, increasing the risk of secret exposure, unintended file access, or outbound requests to third-party services.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

For a markdown file, trigger phrases should be specific enough to avoid unintended invocation. "内容分发" is a generic content/marketing term that could match many unrelated workflows, and the file does not provide exclusion examples for this phrase.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The condition covers any user wanting to turn a reference case into original content and asking for scripts, prompts, code, or task lists. This describes a wide range of general creative assistance requests and does not clearly distinguish when this specialized multi-platform video adaptation skill should activate.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The phrase covering e-commerce, advertising, marketing, livestream commerce, recommendation content, short drama, manga drama, or social media delivery is very broad and context-based rather than trigger-specific. Without explicit constraints, it risks invoking the skill for many general marketing requests not requiring this workflow.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The default prompt hard-codes Chinese output regardless of user preference, which can cause the system to ignore user language expectations or downstream policy/context requirements. While this is not a direct code-execution issue, it can produce misleading or unusable outputs, reduce transparency for non-Chinese-speaking operators, and increase the chance that users approve actions they do not fully understand.

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 tightly scoped triggers, which can cause the agent to invoke this capability in situations the user did not clearly intend. Because this skill can generate production-ready workflows and runnable AI-HIVE commands for multi-platform content generation/distribution, accidental activation could lead to overbroad automation, unintended third-party API use, or content operations being initiated without sufficient user confirmation.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The manifest describes a skill that receives source video and platform/region/account inputs, then generates platform-specific adaptation outputs and interacts with AI-HIVE OpenAPI for configuration lookup, upload, routing, polling, and download. This file only parses a few CLI fields and writes a deterministic local JSON draft brief, with no video handling, no platform-specific variant generation, and no AI-HIVE API interaction.

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 is presented as a specialized multi-platform video adaptation skill, but the implementation and even the top-level docstring expose a general-purpose AI client covering chat, image generation, model enumeration, uploads, and account-related operations. This capability mismatch increases attack surface and breaks least-privilege expectations, making it easier for a caller or downstream agent to use the skill for purposes outside the reviewed scope.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The module docstring explicitly describes the script as a general AI capability invocation tool, which conflicts with the declared identity of a narrowly scoped video-reversion skill. This discrepancy is dangerous because operators and reviewers may trust the manifest scope while the code actually supports much broader operations.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The file-level description and nearly all user-facing prompts are written only in Chinese, and the skill presents itself as a general-purpose tool rather than a region-specific or Chinese-only skill. Under the policy, language constraints should be opt-in or clearly justified; here no user choice or locale justification is provided.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill can call a user-info endpoint that may reveal account identity and wallet balance, which is unrelated to the stated task of multi-platform video reversion. In an agent setting, unnecessary access to billing/account metadata creates avoidable data exposure and can aid reconnaissance about the linked AI service account.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

Generic text chat is broader than the declared function of adapting a source video into platform-specific variants. In a skill ecosystem, this enables prompt-based use of the same API key for unrelated tasks, undermining policy scoping and making review of intended behavior less reliable.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

Standalone image generation is not required for the skill's stated role and broadens the set of actions that can be performed with the configured API credentials. This unnecessary capability expansion increases the chance of misuse, policy bypass, or accidental cost-incurring actions outside the approved workflow.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

This Python file embeds Chinese-only natural-language identifiers in the module docstring, CLI description, and generated skill metadata. Because the file provides no user choice of language and does not document that the skill is intentionally region-specific, it can violate the policy against forcing a specific language or locale without opt-in.

Content

No source excerpt is available for this finding.