Back to skill

Security audit

家居商品空间植入效果图|AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

The skill appears to support a legitimate AI-HIVE furniture-image workflow, but its API client handles credentials and network destinations too broadly for automatic trust.

Install only if you are comfortable using AI-HIVE with an API key and uploading selected media to its service. Keep the default official base URL, do not use untrusted AI_HIVE_BASE_URL or --base-url values, review generated commands before running them, and protect ~/.ai-hive/config.json because it stores a billable API key.

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

Warning
Location
scripts/imagegen.py:94
Finding

Configurable and Server-Supplied Network Destinations Enable API Key Disclosure and SSRF

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Note
Location
scripts/imagegen.py:4
Finding

Runtime Dependency Is Installed Without Version or Integrity Pinning

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • 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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (12)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The skill claims AI-HIVE OpenAPI integration, pricing/config queries, reference uploads, routing strategies, async polling, downloads, and image-generation workflow behavior, but the file only contains descriptive text and example commands without implementing or verifying those controls. This mismatch is dangerous because users or orchestrators may trust the skill to perform cost-sensitive or policy-sensitive operations safely when no actual enforcement, validation, or integration exists.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding

The skill presents executable workflow examples that rely on environment variables, filesystem access, network calls, and shell execution, but it does not declare any tool scope or permissions boundary. In an agent environment, this can cause overbroad capability exposure or unclear operator expectations, increasing the chance that the skill is run with more privilege than necessary.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The invocation list includes 家居电商, which is a broad domain label rather than a specific task-oriented trigger phrase. Because the skill activates when a user merely 'mentions' this term, it could overlap with many ordinary discussions about home-furnishing ecommerce that do not actually request this skill.

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 regardless of the user's language preference. This can misalign the model response with user intent, obscure safety or compliance details for non-Chinese-speaking users, and make review of generated workflow steps harder when the skill is implicitly invoked.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill enables implicit invocation without narrowing conditions, examples, or user-confirmation gates. Because this skill can trigger AI-HIVE workflow generation and API-oriented actions for commercial image synthesis, broad auto-activation increases the chance the agent invokes it in unintended contexts, causing unnecessary external processing, cost, or policy-sensitive image manipulation workflows.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The file-level docstring and argparse description specify the skill entirely in Chinese, presenting a fixed language/locale with no option for user selection or opt-in. Under the policy, forcing a specific language without user choice is a natural-language policy violation unless the locale restriction is explicitly justified, which is not present here.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The module presents itself as a general-purpose AI client even though the packaged skill is supposed to be narrowly scoped to furniture room-placement image generation. This capability/description mismatch is dangerous because it signals that broader functions may be intentionally retained behind a restricted-looking skill wrapper, increasing the chance of policy bypass, unintended use, or operator confusion about what the skill can actually do.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The module docstring and subsequent CLI help and status messages are written entirely in Chinese, which imposes a specific language on all users. The file does not offer any language selection, fallback, or state that the skill is intentionally limited to a Chinese-speaking or region-specific audience.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The file includes text-chat and video-generation API methods that exceed the declared scope of a furniture image-placement skill. In a constrained skill ecosystem, this hidden breadth enables capability creep and potential misuse by callers who can invoke non-approved AI functions through a trusted packaging context.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest focuses on generating furniture-in-room marketing previews, uploading references, polling tasks, and downloading outputs. The additional ability to fetch user info and wallet balance is an account-management capability not needed to fulfill that purpose.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The manifest describes a purpose-built furniture placement image workflow, but the code includes general model listing and lookup across TEXT, IMAGE, and VIDEO model types. Broad discovery of platform models is a generic platform capability rather than something inherent to this skill's stated function.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

The generate subcommand help text claims it can 'generate or edit' images. In implementation, skill_generate simply validates inputs and forwards everything to cmd_image, which always calls the generation/image endpoint with prompt plus optional reference images; there is no distinct edit-mode behavior or edit-specific operation documented in code.

Content

No source excerpt is available for this finding.