Back to skill

Security audit

Build SDF Liquid Glass

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent React UI helper for adding Vaso-based liquid-glass components, with one dependency-pinning caution but no evidence of hidden or malicious behavior.

Before installing, pin `vaso` to a reviewed version, keep the project lockfile, and install dependencies in a normal least-privileged development environment. The skill itself appears limited to UI implementation guidance and reusable React/CSS assets.

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:12
Finding

Unpinned Third-Party Dependency Installation

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:12; related usage at assets/react/SDFLiquidGlass.tsx:2
Vulnerability Type: Unpinned third-party dependency
Risk Level: Medium

Vulnerable code snippets:

SKILL.md:12

markdown
2. Install `vaso` with the repository's package manager.

assets/react/SDFLiquidGlass.tsx:2

tsx
import { Vaso } from 'vaso'

Technical Analysis

The skill instructs the agent to install vaso without specifying an audited version, immutable source commit, lockfile requirement, or integrity hash. The bundled React component subsequently imports and executes the installed package.

Package managers may therefore resolve whichever package version currently satisfies the unconstrained request. If the package, a transitive dependency, or its publishing account is compromised, an attacker could introduce malicious lifecycle scripts or browser-side code after this skill has been reviewed. The linked upstream repository in references/tuning.md documents the intended project but does not ensure that installation resolves to a verified artifact from that repository.

This is a supply-chain weakness rather than evidence that the current vaso package is malicious.

Attack Path

  1. An attacker compromises the package publisher, a transitive dependency, or the relevant registry artifact.
  2. The attacker publishes a malicious version under the expected package name.
  3. A user invokes the skill and follows its instruction to install vaso without an exact version or verified integrity value.
  4. The package manager resolves the attacker-controlled release.
  5. Malicious code executes through an installation lifecycle script or when the React application imports and runs the package.

Impact Assessment

Install-time code could execute with the permissions of the user or automation account running the package manager. Depending on that environment, the affecte ...[truncated 517 chars]

Remediation
View remediation

Remediation Suggestions

  1. Replace the unconstrained installation instruction with an exact, reviewed package version, such as vaso@<audited-version>.
  2. Commit and enforce the package manager's lockfile so direct and transitive dependency versions remain reproducible.
  3. Verify registry provenance, package ownership, and integrity metadata before installation.
  4. Pin to an immutable, reviewed commit when sourcing directly from a repository.
  5. Run dependency installation with lifecycle scripts disabled where compatible with the package and build process.
  6. Use automated dependency scanning, but require review and testing before accepting version updates.
  7. Install dependencies under a least-privileged account and avoid exposing unnecessary credentials to package-manager processes.
Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep