deAPI - AI Media Generation Toolkit

• Outcome: Pass
• Latest audit: May 29, 2026 · 8:19 PM UTC
• Version: 1.0.0

Overview

This is a coherent deAPI media-processing skill, but users should understand that their selected prompts, text, files, and media URLs are sent to deAPI for cloud processing.

Install only if you are comfortable sending selected prompts, text, image/audio/video URLs, and uploaded media to deAPI. Use a dedicated API key stored as an environment variable, avoid secrets or regulated data unless deAPI is approved for that use, verify URLs before processing, and clean up temporary downloaded images when handling sensitive files.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (27)

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill advertises media processing features but does not clearly disclose that user-provided audio, video, images, and text will be sent to an external third-party service. This can cause users to unknowingly transmit sensitive or regulated data off-platform, creating privacy, compliance, and data-handling risks.

Missing User Warnings

Low

• Confidence: 86% confidence
• Finding: The setup section instructs users to configure an API key but does not warn them to protect the credential or avoid exposing it in prompts, logs, screenshots, repositories, or client-side code. This increases the chance of accidental credential leakage, which could allow unauthorized use of the deAPI account and depletion of credits or access to associated data.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The skill sends user-provided text to a third-party API for embedding generation but does not clearly warn the user that their input will leave the local environment. This creates a privacy and data-handling risk, especially if users paste sensitive, proprietary, or regulated content assuming the operation is local.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The skill explicitly sends user-supplied text to a third-party TTS service but does not warn users that their content will leave the local environment. This is a real privacy/security issue because users may provide sensitive text assuming local processing, and the skill also transmits it along with an API-backed authenticated request to an external vendor.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The skill instructs sending user-supplied prompts directly to a third-party API using an API key, but it does not warn the user that their content will leave the local environment. This creates a real privacy and consent issue because prompts may contain sensitive or proprietary information, and the external transmission is central to the skill's behavior.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill sends user-provided prompts and image URLs to a third-party API without any explicit user-facing disclosure or consent step. In an agent context, users may reasonably assume inputs stay local, so silent transmission can expose sensitive text, private links, or proprietary media references to an external service.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The skill instructs the agent to download a user-supplied remote image and then upload it to a third-party OCR service without an explicit warning or consent step about external network transfer and third-party processing. This creates a real privacy and data-handling risk, especially if users provide sensitive screenshots, IDs, receipts, or documents and do not realize the content leaves the local environment.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: If a user supplies an image URL, the skill instructs the agent to download that image and then upload it to deAPI without any explicit user-facing disclosure that the content will be transferred to a third-party service. This creates a privacy and data-handling risk, especially if the image contains personal, sensitive, or proprietary information the user may not expect to leave their environment.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill explicitly instructs sending user-supplied media URLs and an authorization token to a third-party API, but provides no privacy notice, consent step, or data-handling warning. Because transcription content may contain sensitive audio/video and the service receives both the target media reference and account-scoped credentials, this creates a real data-sharing and privacy risk even if the service is legitimate.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The skill instructs the agent to download user-supplied image URLs locally and then upload the resulting image content to deAPI, but it never tells the user that their supplied media will be transmitted to a third-party service. This creates a real privacy and data-handling risk, especially if users provide sensitive or private images and do not realize they are being sent off-platform.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: If the user supplies a URL, the skill downloads that remote resource to a temporary local file and then uploads it to a third-party API, but it provides no user-facing notice about either transmission step or temporary local storage. This can expose sensitive images, create privacy/compliance issues, and enables server-side fetching of attacker-controlled URLs without documented validation or restrictions.

External Transmission

Medium

• Category: Data Exfiltration

Content

## Step 2: Send request

```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/txt2embedding" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

• Confidence: 91% confidence
• Finding: The hardcoded remote endpoint indicates the skill depends on a third-party API, creating an external data exposure path. While normal for this type of integration, the danger is elevated because the command encourages direct submission of arbitrary text without clarifying trust boundaries or data governance.

External Transmission

Medium

• Category: Data Exfiltration

Content

## Step 2: Send request

```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/txt2embedding" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

• Confidence: 91% confidence
• Finding: The hardcoded remote endpoint indicates the skill depends on a third-party API, creating an external data exposure path. While normal for this type of integration, the danger is elevated because the command encourages direct submission of arbitrary text without clarifying trust boundaries or data governance.

External Transmission

Medium

• Category: Data Exfiltration

Content

## Step 2: Send request

```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/txt2audio" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

• Confidence: 95% confidence
• Finding: This command performs an authenticated POST to an external API and includes user-provided text in the request body. In the context of a TTS skill this transmission is expected functionality, but it is still a true data-exfiltration boundary because any supplied content is disclosed to a third party without an explicit safety/privacy gate.

External Transmission

Medium

• Category: Data Exfiltration

Content

## Step 2: Send request

```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/txt2audio" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

• Confidence: 95% confidence
• Finding: This command performs an authenticated POST to an external API and includes user-provided text in the request body. In the context of a TTS skill this transmission is expected functionality, but it is still a true data-exfiltration boundary because any supplied content is disclosed to a third party without an explicit safety/privacy gate.

External Transmission

Medium

• Category: Data Exfiltration

Content

## Step 2: Send request

```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/txt2img" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

• Confidence: 87% confidence
• Finding: This command performs a POST request to an external service and includes user-provided prompt data plus authentication credentials for processing. In an agent skill context, undisclosed outbound transmission can expose sensitive prompts or metadata to a third party, making this more than a harmless implementation detail.

External Transmission

Medium

• Category: Data Exfiltration

Content

## Step 2: Send request

```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/txt2img" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

• Confidence: 87% confidence
• Finding: This command performs a POST request to an external service and includes user-provided prompt data plus authentication credentials for processing. In an agent skill context, undisclosed outbound transmission can expose sensitive prompts or metadata to a third party, making this more than a harmless implementation detail.

External Transmission

Medium

• Category: Data Exfiltration

Content

**For Text-to-Video:**
```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/txt2video" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

• Confidence: 92% confidence
• Finding: This command transmits the user's prompt to deAPI along with an authorization token, which is expected for the feature but still constitutes external data exfiltration from the local agent boundary. The main risk is privacy and data-governance exposure if users enter confidential prompts or if the receiving service logs or retains submitted content.

External Transmission

Medium

• Category: Data Exfiltration

Content

**For Text-to-Video:**
```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/txt2video" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

• Confidence: 92% confidence
• Finding: This command transmits the user's prompt to deAPI along with an authorization token, which is expected for the feature but still constitutes external data exfiltration from the local agent boundary. The main risk is privacy and data-governance exposure if users enter confidential prompts or if the receiving service logs or retains submitted content.

External Transmission

Medium

• Category: Data Exfiltration

Content

**For Image-to-Video (animation):**
```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/img2video" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{

• Confidence: 93% confidence
• Finding: The image-to-video flow sends a user-supplied image URL to an external API, which may reveal private or signed URLs, internal asset locations, or sensitive media references. In an agent skill, forwarding URLs without warning can unintentionally disclose user resources to a third party.

External Transmission

Medium

• Category: Data Exfiltration

Content

**Note:** This endpoint requires `multipart/form-data` with file upload.

```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/img2txt" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -F "image=@{local_file_path}" \
  -F "model=Nanonets_Ocr_S_F16"

• Confidence: 89% confidence
• Finding: This command sends image content to an external API endpoint for processing, which is expected for a cloud OCR integration but still constitutes data exfiltration outside the agent environment. The risk is heightened because OCR inputs often contain sensitive text, and the skill does not describe minimization, consent, retention, or redaction safeguards.

External Transmission

Medium

• Category: Data Exfiltration

Content

**Note:** This endpoint requires `multipart/form-data` with file upload.

```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/img-rmbg" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -F "image=@{local_file_path}" \
  -F "model=Ben2"

• Confidence: 88% confidence
• Finding: The skill sends user-provided image content to an external API endpoint, which is an actual external data transmission. In the context of a media-processing skill this behavior is expected, but it is still security-relevant because files may contain sensitive visual data and metadata, and the skill does not present a clear consent or privacy warning at the point of transfer.

External Transmission

Medium

• Category: Data Exfiltration

Content

**For video:**
```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/vid2txt" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"video_url": "$ARGUMENTS", "include_ts": true, "model": "WhisperLargeV3"}'

• Confidence: 92% confidence
• Finding: This command sends user-provided video URLs to an external service along with a bearer API key, which is a genuine external data transmission. In the context of a transcription skill this behavior is expected, but it is still security-relevant because potentially sensitive media and account credentials are involved and there are no safeguards around consent, URL validation, or disclosure.

External Transmission

Medium

• Category: Data Exfiltration

Content

**For video:**
```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/vid2txt" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"video_url": "$ARGUMENTS", "include_ts": true, "model": "WhisperLargeV3"}'

• Confidence: 92% confidence
• Finding: This command sends user-provided video URLs to an external service along with a bearer API key, which is a genuine external data transmission. In the context of a transcription skill this behavior is expected, but it is still security-relevant because potentially sensitive media and account credentials are involved and there are no safeguards around consent, URL validation, or disclosure.

External Transmission

Medium

• Category: Data Exfiltration

Content

**For audio:**
```bash
curl -s -X POST "https://api.deapi.ai/api/v1/client/aud2txt" \
  -H "Authorization: Bearer $DEAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"audio_url": "$ARGUMENTS", "include_ts": true, "model": "WhisperLargeV3"}'

• Confidence: 91% confidence
• Finding: This audio transcription path also transmits user-provided media URLs and a bearer token to a third-party service. That is functionally necessary for the skill, but still a true security/privacy concern because audio may contain sensitive information and the instructions do not mention consent, handling limits, or restrictions on what URLs may be sent.

VirusTotal

66/66 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.