Back to skill

Security audit

观众评论续写短剧|AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly coherent with AI-HIVE media generation, but its shipped clients handle API keys and media transfers without enough destination and file-safety controls.

Review this skill before installing if you will use real AI-HIVE credentials or sensitive media. Use only the official HTTPS AI-HIVE API endpoint, avoid custom base URLs unless you control them, upload only files you are willing to send to the provider, and prefer --no-download or a dedicated output directory until URL and filename validation are tightened.

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 (4)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/imagegen.py:91
Finding

Bearer Credential Disclosure Through an Unrestricted API Base URL

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/imagegen.py:280
Finding

Unvalidated Server-Provided URLs Enable Arbitrary Outbound Requests and Unintended Media Transfer

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/imagegen.py:318
Finding

Server-Controlled Output Names Permit Directory Traversal and Unbounded File Downloads

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Note
Location
scripts/imagegen.py:7
Finding

Unpinned Third-Party Dependency Installation

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 (19)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The manifest frames the skill as ingesting prior episodes, comments, and story constraints to produce creative planning outputs, but the observed capabilities include unrelated generic chat, image/video generation, account queries, browser/API-key setup, and other undeclared actions. Such under-disclosure breaks the trust boundary: an apparently safe planning skill may actually trigger external tooling, handle credentials, or interact with account resources beyond what the user reasonably authorized.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The manifest frames the skill as ingesting prior episodes, comments, and story constraints to produce creative planning outputs, but the observed capabilities include unrelated generic chat, image/video generation, account queries, browser/API-key setup, and other undeclared actions. Such under-disclosure breaks the trust boundary: an apparently safe planning skill may actually trigger external tooling, handle credentials, or interact with account resources beyond what the user reasonably authorized.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The manifest frames the skill as ingesting prior episodes, comments, and story constraints to produce creative planning outputs, but the observed capabilities include unrelated generic chat, image/video generation, account queries, browser/API-key setup, and other undeclared actions. Such under-disclosure breaks the trust boundary: an apparently safe planning skill may actually trigger external tooling, handle credentials, or interact with account resources beyond what the user reasonably authorized.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding

The skill advertises executable workflows that use environment variables, local file access, network calls, and shell commands, but it declares no explicit tool scope or permissions boundary. In agent environments, that omission can cause overbroad tool access, making it easier for prompts or downstream code paths to read secrets, modify files, or perform unintended external actions without a clearly constrained authorization model.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The activation list includes generic phrases such as “观众共创”, “互动剧情”, “连载短剧”, and broader intent descriptions like converting a reference case into original content or supporting multiple marketing/social scenarios. These triggers are not tightly constrained to this specific skill and may cause unintended invocation for users asking about general storytelling, co-creation, or marketing help.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The natural-language instruction in the manifest says to produce a "production-ready Chinese workflow," which imposes a specific language choice by default. The policy allows flagging locale or language constraints when the skill does not offer an explicit user choice or opt-in.

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 presents itself as a skill-specific image tool, but actually includes broad OpenAPI capabilities such as user-info lookup, generic chat, model enumeration, media upload, and video generation. This scope mismatch increases the chance that a caller or orchestrator grants the skill more authority than intended, enabling unintended data access or off-purpose actions through a supposedly constrained skill interface.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This Python skill uses Chinese-only natural-language descriptions and user-facing instructions, beginning with the module docstring and continuing throughout the CLI help text. Because the file does not offer any user language selection or document that it is intentionally limited to a Chinese-speaking context, it violates the language/locale policy requirement.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

Providing unrestricted text chat in a skill advertised as a constrained short-drama/image workflow allows the skill to be repurposed beyond its declared boundaries. In an agent environment, that can bypass policy assumptions, leak uploaded media into unrelated prompts, or let users invoke a general-purpose model through a narrowly approved tool.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The file presents itself as a specialized short-drama/video skill, but the implementation is a broad AI Hive wrapper exposing chat, image generation, model listing, uploads, and task polling. This scope expansion defeats user and platform expectations, increasing the chance the skill is repurposed for unrelated or policy-violating tasks without additional review.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This Python file embeds its user-facing description, help text, prompts, and status messages almost entirely in Chinese, beginning with the module docstring and continuing throughout the CLI UX. The policy allows locale constraints only when users are given a choice or the restriction is clearly justified as region-specific; here, the skill does not offer a language option or state that Chinese-only operation is required.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The unrestricted text chat command allows arbitrary prompting against the provider, which is materially broader than the manifest's narrow short-drama continuation use case. In a skill ecosystem, this kind of hidden general-purpose capability can be abused to bypass intended task scoping, policy controls, or user expectations about what the skill is allowed to do.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The manifest description and title define usage entirely in Chinese and frame invocation around Chinese-language requests, but do not explicitly state whether other languages are supported or that Chinese is required by user choice. Under the language/locale policy, forcing a specific language without opt-in can be a natural-language policy concern unless clearly justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The script hard-codes Chinese-language skill metadata and descriptions in user-facing strings, including the module docstring, argparse description, skill name, and search aliases. This imposes a specific language/locale without any opt-in or alternative, which matches the natural-language locale policy concern.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

Exposing account/profile and wallet balance retrieval is unnecessary for a comment-driven short-drama image skill and expands the accessible data surface. Even if the data is limited to the authenticated user, it can reveal billing or account metadata that an invoking agent or downstream logs should not access under least-privilege principles.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill can call the user-info endpoint and the CLI advertises querying wallet balance, which is not part of generating comment-driven short-drama proposals, storyboards, or video assets. This capability is not justified by the manifest's stated purpose and looks like leftover generic API tooling.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

Generic image generation is outside the declared video-centric, comment-driven short-drama workflow and creates unnecessary capability expansion. While not inherently dangerous, it increases misuse potential and weakens trust boundaries by allowing unrelated content generation through a specialized skill wrapper.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The code automatically creates the output directory and downloads generated files into it when task polling completes. Although an --output-dir argument exists, there is no confirmation prompt or explicit warning near the automatic write behavior to disclose that running generate will persist files locally unless --no-download is used.

Content

No source excerpt is available for this finding.