skills/timothyvang/claude-code-qc/qc/Gen Agent Trust Hub

qc

Pass

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The installation instructions and the included install.sh script utilize a piped execution pattern (curl | bash) to fetch and deploy the skill from the author's GitHub repository (github.com/TimothyVang). This is the intended installation method for the utility.
  • [EXTERNAL_DOWNLOADS]: The skill automates the download and installation of the official GitHub CLI (gh) and its associated GPG keys from cli.github.com and GitHub's package registries across multiple operating systems.
  • [COMMAND_EXECUTION]: The skill executes shell commands for git repository management (git add, git commit, git push) and uses system package managers (apt, brew, dnf, pacman, apk, winget) to install dependencies. Commands involving sudo are included for Linux package installation, with instructions that explicitly respect user prompts.
  • [PROMPT_INJECTION]: The skill contains an Indirect Prompt Injection surface (Category 8) because it is designed to follow instructions found within a repository's local files.
  • Ingestion points: Processes content from CLAUDE.md, CONTRIBUTING.md, .gitmessage, and recent git logs to determine commit formatting.
  • Boundary markers: Absent; the instructions tell the agent to follow project-specific rules discovered in these files exactly.
  • Capability inventory: Includes file system access via git and the ability to execute shell commands and installation scripts.
  • Sanitization: The skill does not perform validation or escaping of the content read from the project configuration files before using them to guide its behavior.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 20, 2026, 03:22 AM
Security Audit — agent-trust-hub — qc