Back to skill

Security audit

Build WebGL Liquid Glass

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent frontend integration helper for adding WebGL liquid-glass UI effects, with a normal but real dependency-installation risk users should review.

Before installing, confirm that adding liquid-gl is appropriate for the project, pin or lock the dependency version, and review package changes like you would for any frontend dependency. The bundled code itself is narrowly focused on rendering and fallback behavior.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:10
Finding

Unpinned Third-Party Dependency Installation

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 10–13
Vulnerability Type: Unpinned third-party dependency installation
Risk Level: Medium

Vulnerable Code

md
## Workflow

1. Inspect the existing framework, styling conventions, target background, and browser support.
2. Install `liquid-gl` with the repository's package manager.
3. Adapt `assets/react/WebGLLiquidGlass.tsx`, `webgl-liquid-glass.css`, and `liquid-gl.d.ts` for React projects. Preserve the same DOM layering in other frameworks.

Technical Analysis

The workflow directs the agent to install liquid-gl without specifying an exact audited version, lockfile requirement, integrity hash, or trusted package source. Consequently, the package manager may resolve a release that differs from the version originally reviewed.

Third-party package installation can also invoke package lifecycle scripts in the context of the developer or build environment. Although references/tuning.md identifies a GitHub repository as the project source, the installation instruction does not bind the installed package to that repository, a verified registry identity, or a reviewed commit.

This creates a software supply-chain risk if the package name is substituted, the publishing account or registry artifact is compromised, or a future release introduces malicious behavior.

Attack Path

  1. An attacker compromises the package publisher, registry artifact, or another component in the resolved dependency tree.
  2. The attacker publishes a malicious release under the dependency identity that the package manager will resolve.
  3. An agent follows the Skill workflow and installs liquid-gl without an exact version or integrity constraint.
  4. The package manager retrieves the attacker-controlled release and may execute its lifecycle scripts during installation.
  5. Malicious code then runs in the installation or build context. If incorporated into the application bundle, it may also execute in ...[truncated 630 chars]
Remediation
View remediation

Remediation Suggestions

  1. Replace the generic installation instruction with an exact, audited package version, such as liquid-gl@X.Y.Z.
  2. Require installation from the verified official registry identity and confirm that it corresponds to the documented upstream repository.
  3. Commit and enforce the package-manager lockfile, including integrity metadata, in source control.
  4. Use reproducible installation commands such as npm ci rather than unconstrained dependency resolution in automated environments.
  5. Review the package and its complete transitive dependency tree before approving upgrades.
  6. Disable dependency lifecycle scripts where operationally feasible, or explicitly inspect required scripts before allowing them to run.
  7. Use automated dependency monitoring and require security review for lockfile or package-version changes.
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 14)May include surrounding context.

md
3. Adapt `assets/react/WebGLLiquidGlass.tsx`, `webgl-liquid-glass.css`, and `liquid-gl.d.ts` for React projects. Preserve the same DOM layering in other framewo

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill description contains broad default-selection language such as using this as the default skill and preferring it over alternatives, but it does not define clear activation boundaries. In an agent setting, this can cause over-invocation of the skill outside its safest or most appropriate context, reducing user choice and increasing the chance of unnecessary package installation or unsuitable architectural decisions.

Content

No source excerpt is available for this finding.