Back to skill

Security audit

爆款视频黄金3秒钩子分析|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 handles a persistent API key, can run billable upload/generation/download workflows, and accepts unvalidated custom API endpoints.

Install only if you are comfortable giving the skill an AI-HIVE API key and using it for media generation workflows that may cost money. Do not use custom base URLs unless you fully trust them, review prompts and routing before generation, use --no-download when you only want status, and keep API keys out of logs, screenshots, and shared projects.

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:89
Finding

AI-HIVE API Key Disclosure Through a User-Controlled Base URL

Content
View full analysis
Remediation
View remediation
str: parsed = urlparse(value) if parsed.scheme != "https": raise SystemExit("The API base URL must use HTTPS.") if parsed.hostname not in ALLOWED_API_HOSTS: raise SystemExit("The API hostname is not trusted.") if parsed.username or parsed.password or parsed.fragment: raise SystemExit("The API base URL contains prohibited components.") return value.rstrip("/") ``` ]]>

T08 · Insecure Dependencies

Note
Location
scripts/videogen.py:7
Finding

Unpinned Runtime Dependency Installation Guidance

Content
View full analysis
Remediation
View remediation
\ --hash=sha256: ``` ]]>
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
97% confidence
Finding

The skill's stated purpose is hook/opening analysis, but its effective behavior appears to include generic chat, image/video generation, model enumeration, pricing snapshots, wallet/account info retrieval, API-key setup, uploads, polling, and downloads without implementing the promised analysis logic. This makes the skill more dangerous in context because it can act as a broad external-service operator under a specialized branding, increasing the risk of credential misuse, privacy exposure, and unanticipated charges.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The skill's stated purpose is hook/opening analysis, but its effective behavior appears to include generic chat, image/video generation, model enumeration, pricing snapshots, wallet/account info retrieval, API-key setup, uploads, polling, and downloads without implementing the promised analysis logic. This makes the skill more dangerous in context because it can act as a broad external-service operator under a specialized branding, increasing the risk of credential misuse, privacy exposure, and unanticipated charges.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The skill's stated purpose is hook/opening analysis, but its effective behavior appears to include generic chat, image/video generation, model enumeration, pricing snapshots, wallet/account info retrieval, API-key setup, uploads, polling, and downloads without implementing the promised analysis logic. This makes the skill more dangerous in context because it can act as a broad external-service operator under a specialized branding, increasing the risk of credential misuse, privacy exposure, and unanticipated charges.

Content

No source excerpt is available for this finding.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 77)May include surrounding context.

md
python3 scripts/videogen.py init --skill-name viral-video-hook-analyzer-ai-hive

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 79)May include surrounding context.

md
python3 scripts/videogen.py init --skill-name viral-video-hook-analyzer-ai-hive

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The implementation is a broad AI-Hive media generation client with chat, model discovery, upload, download, and video/image generation, while the skill is advertised as a narrowly scoped hook-analysis tool. This capability mismatch is dangerous because it expands what the agent can do beyond user expectations and declared purpose, increasing the chance of unauthorized media generation, billing consumption, and misuse under misleading branding.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding

The skill documents runnable commands that use environment variables, filesystem access, shell execution, and networked API calls, but it declares no explicit tool scope or permission boundaries. In an agent environment, this increases the chance the skill is invoked with broader capabilities than intended, enabling unintended file access, outbound requests, or execution paths without clear user or platform review.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The trigger phrases include broad, everyday marketing terms such as '开场文案', '完播率', and '爆款视频' without enough narrowing context, making accidental or overly broad activation more likely. In combination with the skill's undeclared powerful capabilities, this increases the chance it is selected in conversations where users did not intend to authorize networked, file, or generation workflows.

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 and a specific workflow transformation regardless of the user's language or preference. While not a direct code-execution flaw, it can cause unintended behavior, reduce user comprehension, and increase the likelihood of the skill being silently applied in contexts where the user did not ask for Chinese or for production-ready command generation.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill enables implicit invocation without defining narrow trigger conditions, exclusions, or confirmation gates. That increases the chance the agent will auto-select this skill in loosely related conversations and inject its default workflow behavior, which is especially risky here because the skill steers users into runnable AI-HIVE commands and external workflow execution.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The docstring and argument parser description specify the skill entirely in Chinese, and the generated metadata also fixes the skill identity and artifacts to Chinese-language labels. There is no indication that users can choose another language or that the skill is intentionally limited to a Chinese-only regional context, which can violate language/locale policy requirements.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest says this skill is focused on first-screen hook analysis and delivering hook judgments plus 10 opening concepts. However, the generated blueprint declares artifacts including '差异化复刻方案', '逐镜脚本', and '生成提示词和验收清单', which move into content replication strategy and end-to-end production planning rather than narrow hook analysis.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The manifest describes a skill focused on analyzing the first three seconds of videos and, when generation is needed, routing through AI-HIVE OpenAPI workflows. This script instead implements standalone local media editing operations such as trim, aspect conversion, mute, loudness normalization, and concatenation, which are production/editing capabilities not justified by a hook-analysis skill's stated purpose.

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:

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The top-level docstring and all user-facing guidance in this file are written exclusively in Chinese, presenting the skill as a general-purpose tool rather than a region-specific or Chinese-only skill. Under the policy, forcing a specific language without user opt-in is a natural-language policy concern.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The code exposes user-info and model-listing operations that reveal wallet/account details and available backend capabilities unrelated to first-screen hook analysis. In a skill presented as a focused marketing-analysis tool, this creates unnecessary privilege and information exposure that could aid reconnaissance or lead to unintended disclosure of account metadata.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

Successful generation results are automatically downloaded and written to a default directory unless the caller already knows to pass --no-download. This creates an unprompted side effect on the local filesystem and can surprise users, consume disk space, or store unreviewed remote content locally in a skill whose branding emphasizes analysis rather than file-writing behavior.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The embedded skill configuration brands the tool as '爆款视频黄金3秒钩子分析', but the parser/description and behavior implement a generic '裸接口视频 Skill' centered on media generation. This inconsistency is a security problem because it can mislead operators and users about the actions the skill may take, undermining informed consent and enabling covert use of broader generation features.

Content

No source excerpt is available for this finding.