Back to skill

Security audit

varg-occasion-video

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill suspicious because it reports that personal photos and occasion details may be sent to hosted AI services without a clear privacy disclosure or approval step; no malicious intent is established.

Before installing, consider whether you are comfortable sending personal photos and event details to the named hosted services. Check their privacy and retention terms, and review the installer and repository source because they are not pinned to fixed versions. Scanner coverage was partial due to an unresolved local reference, so the results do not establish that every referenced item was assessed.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (8)

The photo-generation workflow lacks a privacy warning about sending personal photos and occasion details to external services.

Medium
Category
Output Handling
Confidence
92% confidence
Finding

The README identifies varg and several AI services, so external processing is not concealed, but it does not clearly explain that personal photos and event details will be submitted for processing. Because the advertised inputs include identifiable family images and private occasion information, this omission can undermine informed consent.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

The installation command invokes an unpinned npm package and installs skill content from an unpinned repository reference. A compromised or unexpectedly changed release could execute code during installation or introduce malicious agent instructions; however, the finding incorrectly describes this installer as an MCP server.

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 · SKILL.md (reported line 33)May include surrounding context.

md
The occasion sets the rules: an invitation must carry the names, the date and the place without a single wrong
letter; a holiday greeting must be forwardable to a family group; old photos must still look like the people;
a tribute must stay calm. Generation runs on varg: REST `https://api.varg.ai/v2` (the `varg-ai` skill) or the MCP
server `https://mcp.varg.ai/mcp` (`generate_image`, `generate_video`, `generate_music`, `call_tool`, `get_job`,
`estimate_cost`). Quote the price before anything runs.

The workflow sends personal photos and occasion details to external AI services without an explicit privacy disclosure.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

The workflow directs the agent to submit photos and personal event details to hosted generation services without a specific disclosure and confirmation step for that transfer. The existing agreement requirements for animating faces are valuable, but consent to animation or private delivery does not necessarily establish informed consent to third-party processing.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% 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 · SKILL.md (reported line 74)May include surrounding context.

  1. Music and export with scripts/assemble.sh (scales, joins, lays the track, -14 LUFS, H.264 + AAC).
bash
curl -s -X POST https://api.varg.ai/v2/video -H "Authorization: Bearer $VARG_API_KEY" -H "Content-Type: application/json" \
  -d '{"model":"seedance_2_5","duration":8,"aspect_ratio":"9:16","resolution":"720p","audio":true,
       "files":[{"url":"CLOSED_CARD_URL"},{"url":"OPEN_CARD_URL"}],
       "provider_options":{"piapi":{"input":{"mode":"first_last_frames"}}},

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 · SKILL.md (reported line 74)May include surrounding context.

  1. Music and export with scripts/assemble.sh (scales, joins, lays the track, -14 LUFS, H.264 + AAC).
bash
curl -s -X POST https://api.varg.ai/v2/video -H "Authorization: Bearer $VARG_API_KEY" -H "Content-Type: application/json" \
  -d '{"model":"seedance_2_5","duration":8,"aspect_ratio":"9:16","resolution":"720p","audio":true,
       "files":[{"url":"CLOSED_CARD_URL"},{"url":"OPEN_CARD_URL"}],
       "provider_options":{"piapi":{"input":{"mode":"first_last_frames"}}},

The image-generation workflow includes personal event details without warning about external processing.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The hosted image-generation template includes names, a date, and a city without a disclosure or approval step for external processing. These details are relevant to invitations rather than covert collection, but combining them can expose private event information before the user intends to distribute it.

Content

No source excerpt is available for this finding.

The workflow directs use of external AI services with family photos without disclosing third-party data sharing.

Medium
Category
Output Handling
Confidence
93% confidence
Finding

The restoration and animation steps require submitting identifiable family photographs to hosted models without explaining the third-party data transfer. Family agreement and private delivery reduce misuse risks, but neither addresses provider processing, retention, or onward sharing of the original photos.

Content

No source excerpt is available for this finding.