Back to skill

Security audit

Video Generator

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent Remotion video workflow, but it defaults to public tunneling of a local dev server and mutable package execution without enough user control or disclosure.

Review before installing. Use this only in an isolated workspace, avoid exposing the dev server publicly unless you explicitly need it, stop any tunnel immediately after use, and prefer pinned dependencies and lockfiles before running npm or npx commands. Be aware that product/company videos may trigger website scraping and local asset downloads.

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

Error
Location
SKILL.md:25
Finding

Unauthenticated Public Exposure of the Remotion Development Server

Content
View full analysis
&& npm run dev ``` The workflow then requires: ```bash bash skills/cloudflare-tunnel/scripts/tunnel.sh start 3000 ``` It instructs the agent to send the resulting public URL to the user. ### Technical Analysis The default workflow starts Remotion Studio on port 3000 and exposes that development service to the public Internet through a Cloudflare tunnel. The instructions do not require authentication, an access allowlist, explicit user consent, a review of exposed content, or verification that the development server is safe for public deployment. A public tunnel is not necessary for the core functionality of creating or rendering a video. Local preview, authenticated preview, or a rendered static artifact would satisfy the declared purpose with fewer privileges and a smaller attack surface. Consequently, mandatory tunneling exceeds the minimum network exposure required by the Skill. The workflow also does not require the tunnel to be stopped after previewing. The later tunnel-management section documents a stop command, but cleanup is not part of the mandatory workflow. This can leave the service exposed longer than intended. ### Attack Path 1. The Skill creates a video project containing unpublished media, brand assets, screenshots, or other project information. 2. It starts Remotion Studio on port 3000. 3. It creates a public Cloudflare tunnel to that port without authentication or access restrictions. 4. The public URL is disclosed, leaked through logs or conversation history, or otherwise obtained by an unauthorized party. 5. The unauthorized party accesses content and functionality exposed by the development server. 6. If the development server or its dependencies contain exploitable defects, the attacker ...[truncated 946 chars]
Remediation
View remediation

T08 · Insecure Dependencies

Error
Location
SKILL.md:38
Finding

Execution of Mutable and Unpinned Third-Party Packages

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (12)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill instructs the agent to publicly expose a local dev server and share the URL, but provides no warning about privacy, discoverability, authentication, or data exposure. Users may not understand that in-progress source, assets, and any server-accessible functionality are being made reachable from the internet.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The manifest description says the skill should be used for creating videos, short films, commercials, motion graphics, product demos, social media videos, animated explainers, or 'any programmatic video content.' This trigger scope is broad and lacks explicit constraints or negative examples, which increases the risk of unintended invocation for general video-related requests.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill is presented as a video-generation workflow, but its default behavior includes scraping third-party websites and downloading external assets. That is a scope expansion with privacy, legal, and security implications because user requests for videos may unexpectedly trigger outbound requests and ingestion of untrusted remote content.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The skill directs execution of npx remotion without pinning a version, which allows the latest package version to be fetched at runtime. This creates a supply-chain and reproducibility risk: a compromised upstream release or unexpected breaking change could execute unreviewed code in the agent environment.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The documented npx remotion usage in package scripts is unpinned, so execution may resolve to whatever version is current at runtime. That exposes the workflow to supply-chain compromise and non-deterministic behavior in a privileged local development context.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

This second script entry also relies on unpinned npx remotion, meaning build-time code may come from an unreviewed current upstream release. In automation or agent environments, that materially increases supply-chain exposure because package execution happens automatically.

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 video-production skill, but the workflow also instructs exposing a local development server to the public internet and sharing the URL. This materially changes the trust boundary and creates network exposure not implied by the stated capability, which can surprise users and operators.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
86% confidence
Finding

The render command uses unpinned npx remotion, again permitting execution of whatever release is available at the time of use. Even though rendering is a normal function for this skill, the lack of version pinning makes the command a supply-chain risk.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

npx --yes create-video@latest explicitly pulls and executes the latest scaffolding package without review. This is a stronger supply-chain risk than generic npx use because it both fetches and runs new code automatically during project creation.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
88% confidence
Finding

The dev-server startup again depends on unpinned npx remotion, allowing upstream package drift or compromise to affect a process that stays running and is later exposed over a tunnel. The combination of runtime download plus network exposure increases operational risk.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The workflow makes third-party website scraping and asset download mandatory without clearly warning that user-provided URLs and fetched content will be sent to external services and stored locally. This can leak sensitive targets, import malicious or copyrighted content, and create compliance issues unexpectedly.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill includes built-in commands to expose a local Remotion Studio on the public internet via Cloudflare Tunnel, which is more capability than the core video-generation purpose requires. Public exposure of a local dev service can enable unauthorized access, leakage of in-progress assets, and exploitation of vulnerabilities in the exposed service.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.