Back to skill

Security audit

淘宝天猫主图工作室|AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AI-HIVE workflow for Taobao/Tmall main-image production, with some scope and dependency cautions but no evidence of hidden, destructive, or deceptive behavior.

Install only if you intend to use AI-HIVE for Chinese Taobao/Tmall main-image workflows. Review prompts and route choices before any billable generation, use a virtual environment with a pinned requests version if possible, and upload only files you own or are authorized to use.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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)

T08 · Insecure Dependencies

Warning
Location
scripts/imagegen.py:8
Finding

Unpinned Third-Party Dependency Installation

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (14)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared description presents a full e-commerce main-image production skill with operational workflow and external AI-HIVE API interactions for generating and delivering image assets. The supplied code does not implement that workflow. It only parses command-line inputs and writes a draft JSON brief labeled as a visual blueprint. While some field names and aliases align thematically with the declared domain, the code’s actual purpose is much narrower: generating a static project brief scaffold. This is a material mismatch in primary purpose and implemented capabilities.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
83% confidence
Finding

The skill advertises executable commands and operational code paths involving environment variables, local files, network access, and shell usage, but it does not declare any explicit tool scope or permission boundaries. In an agent setting, this can cause over-privileged execution or ambiguous authorization, increasing the chance that a caller or wrapper grants broader capabilities than intended.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
83% confidence
Finding

The description and invocation wording are entirely framed around Chinese trigger phrases and Chinese-platform usage, but there is no explicit statement that the skill is intentionally region- or language-specific, nor any opt-in for language selection. Under the policy, forcing a specific language or locale without user choice or clear justification is a natural-language policy concern.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The activation list includes broad terms such as “电商设计”, “商品点击率”, and scenario-based requests like “稳定交付” across marketing and social contexts. These phrases are not narrowly scoped to this specific Taobao/Tmall main-image workflow and could cause the skill to activate for many ordinary e-commerce or marketing discussions.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The default prompt is broad enough to activate the skill for loosely related requests and immediately steer the conversation into a production workflow with runnable commands. In a skill that can generate external API actions and marketing assets, overbroad invocation increases the chance of unintended execution paths, misuse for policy-sensitive content, or user requests being reframed without explicit confirmation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

Forcing Chinese output regardless of user preference can override user intent and reduce transparency, especially if warnings, consent prompts, or safety caveats are emitted in a language the user did not request. In a workflow that may produce executable commands and compliance-sensitive marketing content, this can impair informed review and increase the risk of user misunderstanding.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The script hardcodes Chinese-language branding and metadata, including the top-level description, argument parser description, skill name, and search aliases, with no option for user language selection or explanation that the skill is region-specific. This creates a natural-language locale policy concern because the skill appears to enforce one language by default.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The file presents itself as a general AI client supporting chat, image, video, model enumeration, upload, and polling, while the declared skill is supposed to be narrowly scoped to Taobao/Tmall main-image studio workflows. This capability mismatch expands what downstream agents or users can do through the skill and weakens policy boundaries, creating a realistic path for misuse outside the approved business purpose.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This code file contains natural-language strings that present the skill as a general-purpose tool, but its user-facing description is fixed to Chinese. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale restriction is explicitly justified.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The chat subcommand exposes unrestricted text and multimodal chat against arbitrary text models, which is unrelated to a specialized Taobao/Tmall main-image production skill. In an agent environment, this can become a covert general-purpose model access path that bypasses intended tool restrictions and permits tasks the skill was not reviewed to handle.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The video-generation command enables arbitrary video creation, including uploads of image, video, audio, and frame controls, far beyond the stated purpose of producing Taobao/Tmall main images. This materially broadens the attack surface and enables unauthorized or policy-evading media generation workflows under the cover of a narrowly described image skill.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The embedded SKILL_CONFIG exposes multiple user-facing fields entirely in Chinese while the skill is otherwise presented as a reusable generic image skill. This enforces a locale/language choice implicitly rather than offering users a choice or clearly constraining the skill to a Chinese-only context.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
82% confidence
Finding

This code writes JSON output to a user-specified path, which can overwrite an existing file, but there is no confirmation prompt, warning message, or explanatory comment/docstring disclosing that behavior. The printed resolved path happens only after the write, so it does not function as a prior warning.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

The top-level documentation says this is a '通用 AI 能力调用工具' for chat, image, video, model query, media upload, and polling. That actively conflicts with the skill's declared developer intent as a domain-specific Taobao/Tmall main-image studio skill rather than a generic AI client.

Content

No source excerpt is available for this finding.