Back to skill

Security audit

think-x-actually-y-video-gen

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed WeryAI video-generation helper with paid API use and approval gates; the main caution is optional local image upload.

Install only if you are comfortable giving the skill a WeryAI API key and paying for confirmed generation jobs. Prefer public HTTPS image URLs; do not approve local file paths unless you have checked the exact file and are willing to upload it to WeryAI. Review the full prompt and parameters before confirming any submit or wait action.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (18)

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared purpose is a narrowly scoped three-beat debunk template, but the skill text reveals broader operational capabilities including general video generation, image and multi-image upload flows, local file handling, and task polling. This mismatch is dangerous because reviewers and users may trust the benign-looking description while the implementation can transmit local data to a third party and perform actions outside the advertised scope.

Content

No source excerpt is available for this finding.

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 16)May include surrounding context.

md
**Dependencies:** `WERYAI_API_KEY` + Node.js 18+. **`scripts/video_gen.js`** + **[`WERYAI_VIDEO_API.md`](resources/WERYAI_VIDEO_API.md)**. **No other Cursor ski

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 20)May include surrounding context.

md
**Dependencies:** `WERYAI_API_KEY` + Node.js 18+. **`scripts/video_gen.js`** + **[`WERYAI_VIDEO_API.md`](resources/WERYAI_VIDEO_API.md)**. **No other Cursor ski

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 27)May include surrounding context.

md
**Dependencies:** `WERYAI_API_KEY` + Node.js 18+. **`scripts/video_gen.js`** + **[`WERYAI_VIDEO_API.md`](resources/WERYAI_VIDEO_API.md)**. **No other Cursor ski

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 28)May include surrounding context.

md
**Dependencies:** `WERYAI_API_KEY` + Node.js 18+. **`scripts/video_gen.js`** + **[`WERYAI_VIDEO_API.md`](resources/WERYAI_VIDEO_API.md)**. **No other Cursor ski

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 29)May include surrounding context.

md
**Dependencies:** `WERYAI_API_KEY` + Node.js 18+. **`scripts/video_gen.js`** + **[`WERYAI_VIDEO_API.md`](resources/WERYAI_VIDEO_API.md)**. **No other Cursor ski

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 73)May include surrounding context.

md
**Dependencies:** `WERYAI_API_KEY` + Node.js 18+. **`scripts/video_gen.js`** + **[`WERYAI_VIDEO_API.md`](resources/WERYAI_VIDEO_API.md)**. **No other Cursor ski

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

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

md
**Dependencies:** `WERYAI_API_KEY` + Node.js 18+. **`scripts/video_gen.js`** + **[`WERYAI_VIDEO_API.md`](resources/WERYAI_VIDEO_API.md)**. **No other Cursor ski

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 75)May include surrounding context.

md
**Dependencies:** `WERYAI_API_KEY` + Node.js 18+. **`scripts/video_gen.js`** + **[`WERYAI_VIDEO_API.md`](resources/WERYAI_VIDEO_API.md)**. **No other Cursor ski

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 81)May include surrounding context.

md
**Dependencies:** `WERYAI_API_KEY` + Node.js 18+. **`scripts/video_gen.js`** + **[`WERYAI_VIDEO_API.md`](resources/WERYAI_VIDEO_API.md)**. **No other Cursor ski

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
94% confidence
Finding

The skill declares network and environment-variable-dependent behavior but does not define an explicit tool/permission scope. That creates an authorization gap: a caller may invoke a skill that can access secrets and make outbound requests without a clearly constrained execution boundary, increasing the chance of overbroad secret exposure or unintended remote actions.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

Automatically translating Chinese user input into English captions without explicit opt-in can alter user intent and cause unauthorized transformation of content before sending it to the remote video API. In multilingual or regulated contexts, this can create privacy, accuracy, and consent issues, especially when the output is published-facing media.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · eval.yaml (reported line 30)May include surrounding context.

yaml
type: llm_rubric
    rubric: |
      SKILL.md must require user-approved X/Y/Z (or proposed in table), three-beat timing scaled to duration, and full expanded prompt before submit.
      SKILL.md must include explicit pre-submit user confirmation (no submit-* or wait without approval) and default async submit with immediate task-id notification and user-chosen status vs blocking wait; CLI reference must show submit-* and status examples.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The document first states that public HTTPS image URLs are mandatory, then later introduces a local-path upload mechanism that effectively bypasses that rule. This inconsistency can mislead operators and downstream agents into assuming stricter safety properties than actually exist, increasing the chance that local files are accessed and uploaded without appropriate review.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This documentation explicitly describes a code path where non-HTTPS image inputs are treated as local filesystem paths, read from disk, and uploaded to an external service. That expands the effective capability of the skill from video generation into local file access and exfiltration, which is security-sensitive and can be abused if an agent passes user- or model-controlled paths.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The skill is presented as a video-generation helper, but the documented behavior includes reading arbitrary local files and sending them to a remote upload endpoint. Even if intended for convenience, this is a privilege expansion that is not clearly justified by the skill's stated purpose and increases the risk of unintended sensitive-file access or exfiltration.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest describes a narrowly scoped skill for creating vertical three-segment 'you think X, actually Y, but Z' videos with timed English captions. However, the code exposes generic commands for text-to-video, image-to-video, multi-image-to-video, model listing, and arbitrary status polling, and accepts free-form prompts without enforcing three beats, vertical format, or caption timing.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

A skill whose stated purpose is to produce a specific debunk-video format would not obviously need a separate capability to enumerate all available WeryAI video models. This is an auxiliary platform-discovery feature that expands capability beyond the stated template-generation intent.

Content

No source excerpt is available for this finding.