Back to skill

Security audit

App界面演示广告|AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

This appears to be a real AI-HIVE video-ad helper, but it gives agents enough authority to upload media, spend credits, and expose API keys without strong built-in safeguards.

Install only if you trust the publisher and will review commands before running them. Do not use a custom base URL unless you fully control it, rotate any key used with an untrusted endpoint, and treat generate/upload commands as potentially billable and privacy-sensitive because selected local media can be uploaded immediately.

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 (2)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/videogen.py:91
Finding

Bearer API Key Can Be Transmitted to an Arbitrary Server

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/videogen.py:514
Finding

Potentially Billable Generation and Media Uploads Lack an Enforced Confirmation Gate

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 (13)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

A second description-behavior mismatch reiterates that the skill claims structured app-promo orchestration and AI-HIVE operations while only evidencing basic local video manipulation/JSON generation. Such overstated capability can cause an orchestrating agent to pass sensitive media or trust the skill for compliance-sensitive creative workflows it cannot actually perform safely.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

A second description-behavior mismatch reiterates that the skill claims structured app-promo orchestration and AI-HIVE operations while only evidencing basic local video manipulation/JSON generation. Such overstated capability can cause an orchestrating agent to pass sensitive media or trust the skill for compliance-sensitive creative workflows it cannot actually perform safely.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding

The skill advertises runnable code that uses environment variables, filesystem access, network calls, and shell-based tooling, but it does not declare any tool scope or permission boundaries. In agent environments, this can lead to overbroad tool access, making accidental secret exposure, arbitrary file access, or unintended external requests more likely if the skill is invoked.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
81% confidence
Finding

The skill content, invocation guidance, and output requirements are entirely presented in Chinese, and no user language preference or opt-in is offered. Under the stated policy, a skill should not impose a specific language unless it explicitly provides a choice or clearly documents a justified regional constraint.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The invocation phrases are broad marketing terms that overlap with normal conversation, increasing the chance the skill is auto-invoked in contexts where the user did not intend to run ad-production logic. Unintended invocation matters here because the skill includes code-oriented and potentially billable/network-capable workflow guidance, which can expand attack surface or trigger unwanted actions.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The default prompt explicitly instructs the assistant to produce a 'production-ready Chinese workflow,' which constrains output language without checking the user's preference. This can override or conflict with user intent, reduce usability for non-Chinese-speaking users, and create prompt-level behavior that is broader than necessary for the task.

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 file is marketed as a narrowly scoped App UI demo ad skill, but the implementation includes broad-purpose AI-Hive client functionality such as chat, model enumeration, account queries, and arbitrary media upload. This creates a scope mismatch that can mislead operators and downstream systems into granting or invoking capabilities beyond the declared purpose, increasing the chance of unauthorized data access or misuse.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The module docstring and all user-facing instructions are written exclusively in Chinese, presenting the skill as a general-purpose tool rather than a region-specific or Chinese-only skill. Under the stated policy, forcing a specific language/locale without user opt-in is a natural-language policy violation.

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 retrieve user/account information and wallet balance even though that capability is not needed to generate App UI demo ads. In a skill ecosystem, unnecessary sensitive capability expansion violates least privilege and may expose account metadata or financial information to prompts, logs, or unauthorized users.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The free-form text chat feature is unrelated to the declared purpose and allows arbitrary interaction with backend models through the same API key and client. That broadens the attack surface, enables unintended data processing, and can be abused to bypass product-level constraints expected for a purpose-limited ad-generation skill.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The parser and help text expose generic model browsing and image/video tooling, reinforcing that the artifact behaves like a general AI utility rather than a constrained App UI ad skill. In context, this makes accidental or intentional misuse more likely because users can discover and invoke broader functions not implied by the skill metadata.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
89% confidence
Finding

The file hard-codes Chinese-language title/branding in the module docstring and argparse description (App界面演示广告|AI-HIVE) with no indication that users can choose another language or locale. This can violate language/locale policy when a skill imposes a specific language experience without explicit opt-in or justification.

Content

No source excerpt is available for this finding.