Back to skill

Security audit

Kie AI Nano Banana Pro 生图

Security checks for vulnerabilities and agentic risk

Overview

The skill is a straightforward Kie AI image-generation helper, with expected third-party API and optional webhook use but no hidden execution or persistence.

Before installing, understand that prompts, reference image URLs, and any configured callback URL are intended to be sent to Kie AI or to the callback endpoint you provide. Use a trusted API key storage mechanism, avoid submitting confidential prompts or private images unless approved, and only configure webhook callbacks to HTTPS endpoints you control and can verify.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (12)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The README promotes webhook callback support and gives a production example sending task results to a user-specified external URL, but it does not clearly warn that generated content and task metadata will be transmitted off-platform. In a skill that handles image-generation requests, this can lead users to unintentionally exfiltrate prompts, outputs, or identifiers to third-party endpoints they do not control or have not secured.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · README.md (reported line 46)May include surrounding context.

生产环境:

text
批量生成产品图,回调到 https://api.myshop.com/callback

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill presents itself as an official image-generation assistant but does not warn users that prompts, reference image URLs, and possibly callback data will be transmitted to an external third-party API. This creates a privacy and informed-consent risk because users may submit sensitive text or images without realizing they are leaving the local assistant context.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger phrases include generic terms like image generation and Google image model requests, which can cause the skill to activate for broad, ordinary user prompts outside its intended scope. Over-broad activation increases the chance that user content is routed into this third-party workflow without clear intent or consent, especially when the skill sends prompts and image URLs externally.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
95% confidence
Finding

The skill is explicitly designed to transmit user-supplied prompts and optional reference image URLs to an external API endpoint. External transmission is expected for this skill's function, but it is still security-relevant because sensitive user data may be exfiltrated to a third party if the skill is triggered unexpectedly or used without adequate disclosure and input validation.

Content

Scanner excerpt · SKILL.md (reported line 55)May include surrounding context.

API 端点:

text
POST https://api.kie.ai/api/v1/jobs/createTask

请求头:

External Transmission

Medium
Category
Data Exfiltration
Confidence
84% confidence
Finding

The example encourages use of a user-specified callback URL, which introduces outbound transmission to arbitrary external infrastructure. If implemented without strict validation, this can enable sending task results or metadata to attacker-controlled endpoints, creating privacy leakage and possible SSRF-style integration risks depending on the surrounding platform.

Content

Scanner excerpt · SKILL.md (reported line 244)May include surrounding context.

示例 3:生产环境回调

用户输入:

批量生成产品图,回调到 https://api.myshop.com/callback

输出:

json

External Transmission

Medium
Category
Data Exfiltration
Confidence
86% confidence
Finding

The concrete callback example shows direct use of an external callback destination, reinforcing a pattern where generated outputs and task metadata may be delivered to third-party endpoints. In context this is part of normal workflow, but it becomes dangerous if arbitrary callback values are accepted without validation, ownership checks, or clear disclosure.

Content

Scanner excerpt · SKILL.md (reported line 250)May include surrounding context.

json
{
  "model": "nano-banana-pro",
  "callBackUrl": "https://api.myshop.com/api/callback",
  "input": {
    "prompt": "Product photography...",
    "aspect_ratio": "1:1",

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger list includes very broad and generic terms such as "kie", "Kie AI", and "nano banana", which can cause the skill to activate in contexts unrelated to the user's intent. Unintended activation can route user prompts to this skill unexpectedly, increasing the chance of accidental API use, privacy issues, or confusing behavior in multi-skill environments.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
85% confidence
Finding

The README instructs users to obtain and configure an API key but provides no guidance on secure storage, rotation, or avoiding disclosure in prompts, logs, or source control. While this is documentation-level rather than active credential theft, it increases the chance of accidental key exposure and unauthorized API use.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

The skill is written entirely around Chinese-language headings, trigger phrases, examples, and output conventions, which effectively imposes a single language experience. Under the policy, a forced language or locale should either be optional for users or explicitly justified as region-specific, neither of which is stated here.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The package metadata contains Chinese-only user-facing text such as the description and keywords, but does not state that the skill is intended specifically for Chinese-speaking users or offer any language choice. Under the policy, forcing a specific language without opt-in or documented locale justification can be a natural-language policy concern.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The display name, description, keywords, and categories contain Chinese-language text, but the manifest does not indicate that users can choose their preferred language or that the skill is intentionally limited to a Chinese-speaking audience. This can violate language/locale policy expectations when a skill implicitly forces one language without user opt-in or justification.

Content

No source excerpt is available for this finding.