impeccable

Fail

Audited by Gen Agent Trust Hub on Sep 22, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONOBFUSCATIONEXTERNAL_DOWNLOADS
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill's primary launcher script (scripts/impeccable) and its Windows equivalent (scripts/impeccable.cmd) are configured to automatically download binary executables from a non-vendor GitHub repository (https://github.com/pbakaus/impeccable/releases) if they are missing locally. These binaries are subsequently executed with full user privileges. This pattern circumvents standard package manager security controls and introduces a supply-chain risk from an unverified external source.
  • [INDIRECT_PROMPT_INJECTION]: The skill maintains a large attack surface for indirect prompt injection as it is instructed to read and analyze various user-controlled files from the project workspace.
  • Ingestion points: The agent reads project documentation (PRODUCT.md, DESIGN.md) and frontend source files (e.g., .tsx, .html, .vue) to generate critiques and new implementations.
  • Boundary markers: The skill instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the ingested project data.
  • Capability inventory: The agent has high-privilege capabilities including filesystem modification, shell command execution via its internal engine, and browser automation.
  • Sanitization: There is no evidence of sanitization or escaping logic applied to project-derived content before it is interpolated into prompts or used to drive code generation.
  • [COMMAND_EXECUTION]: The skill frequently executes shell commands to perform its duties. This includes running its own downloaded engine binary and interacting with system tools such as git, adb (for Android device management), and xcrun (for iOS simulation).
  • [OBFUSCATION]: The skill includes scripts/modern-screenshot.umd.js, which is a large minified JavaScript bundle. This minification obscures the underlying logic and complicates security auditing of the code that the skill injects into browser sessions during design reviews.
  • [EXTERNAL_DOWNLOADS]: The skill fetches external resources (binaries and checksums) from GitHub during its setup process, which is initiated whenever the impeccable context command is called.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Sep 22, 2026, 09:30 AM
Security Audit — agent-trust-hub — impeccable