Back to skill

Security audit

Seedance 2.5 提示词专家|AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly coherent for AI-HIVE video workflows, but its credential and download handling create real review-worthy security risks.

Install only if you are comfortable giving this skill an AI-HIVE API key and access to selected media files. Avoid custom base URLs unless you fully trust the endpoint, review generated commands before running them, prefer --no-download for untrusted tasks, and keep API keys out of logs 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 (4)

T09 · Insecure Skill Coding Practices

Error
Location
scripts/videogen.py:91
Finding

Unrestricted API Endpoint Override Can Disclose the Bearer Token

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
scripts/videogen.py:327
Finding

Server-Controlled Task Metadata Can Escape the Output Directory

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/videogen.py:279
Finding

Unvalidated Remote URLs Permit Media Redirection and Internal Network Requests

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Note
Location
scripts/videogen.py:7
Finding

Unpinned Dependency Installation Produces Non-Reproducible Builds

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
98% confidence
Finding

The declared description promises a comprehensive Seedance 2.5 workflow skill with generation, reference, editing, extension, and AI-HIVE API-backed media task operations. The supplied code does not implement those capabilities. Its actual purpose is narrowly limited to generating a basic JSON brief from command-line arguments. While some fields loosely align with the domain (search aliases, required facts, artifacts), that is only metadata scaffolding and not the described operational behavior. Therefore the description materially overstates the implemented functionality.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

This is a clear description-behavior mismatch. The declared purpose centers on Seedance 2.5 creative assistance and API-backed media-generation workflow support. The supplied code instead implements a narrow local video-editing CLI around ffmpeg/ffprobe. While 'video editing' is mentioned in the description, the code does not provide the described high-level Seedance expertise, prompt creation, reference-media workflow, or any network/API behavior. Its primary purpose is materially different: deterministic local media manipulation rather than Seedance prompt/workflow execution.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding

The skill advertises and demonstrates operations that imply access to environment variables, local files, network services, and shell-executed tooling, but it declares no explicit tool scope or permission boundary. That creates an authorization ambiguity where an agent or reviewer may permit broader capabilities than intended, increasing the risk of secret exposure, unsafe file access, or unintended command/network execution.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The default prompt hard-codes Chinese output regardless of the user's language preference, which can override user intent and reduce transparency about what the system is doing. In a security-sensitive workflow that may generate executable commands and operational guidance, forcing a language without opt-in increases the risk of user misunderstanding, accidental execution, or missed warnings.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill enables implicit invocation while the metadata describes a broad set of loosely bounded use cases, which can cause the agent to trigger this skill for requests the user did not clearly intend. Because the skill can produce runnable AI-HIVE commands and workflow steps, unintended activation could steer conversations into external-service actions, cost-incurring operations, or inappropriate content generation paths without clear user consent.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The script description and embedded content are presented in Chinese-only strings such as the tool name, aliases, and required facts, with no option to select another language or locale. This can violate language/locale policy where skills must not force a specific language without explicit user choice or documented regional scope.

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
91% confidence
Finding

The top-level module docstring says this file is a '通用 AI 能力调用工具' that wraps text chat, image generation, video generation, model querying, media upload, and task polling. However, the skill manifest describes a specialized Seedance 2.5 prompt/video workflow skill; exposing broad generic AI-Hive capabilities in the same file is a semantic mismatch between claimed skill scope and actual implemented behavior.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This code file contains natural-language strings and documentation entirely in Chinese, including the module docstring and user-facing CLI guidance. The skill does not provide an explicit language/locale choice or justify a Chinese-only restriction, which can violate language/locale policy for generally applicable skills.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The file contains general-purpose text chat and image generation capabilities that exceed the declared Seedance 2.5 prompt/video scope. Even though the skill-specific entrypoint currently exposes only generate/task/upload, bundling broader API actions in the same skill increases the risk of capability confusion, accidental exposure in future changes, or repurposing of the skill to perform actions users and reviewers would not expect.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

This code creates parent directories and writes a JSON file to a user-specified path, which is a file-modifying operation. Although expected for a blueprint generator, the file contains no docstring detail, comment, or user-facing message before the write explaining that it will create directories and overwrite/write the output path.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
92% confidence
Finding

The manifest describes a creative workflow assistant for Seedance 2.5 prompting and media generation operations. The additional ability to fetch user info and wallet balance is an account-inspection capability that is not necessary to provide prompt-writing, editing, extension, or media-generation orchestration.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The comment says the browser is opened with a query parameter '为后端归因预留', implying a limited attribution purpose. In practice, the code includes the concrete skill name in the outbound URL, which discloses per-skill usage context to the remote service; that is more specific behavior than the comment suggests.

Content

No source excerpt is available for this finding.