The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The project documentation recommends an installation method that pipes a remote script directly into the shell: 'curl -fsSL "https://raw.githubusercontent.com/Dicklesworthstone/ultimate_bug_scanner/main/install.sh?$(date +%s)" | bash'. This follows a common distribution pattern for developer tools but involves executing remote code locally.
[EXTERNAL_DOWNLOADS]: The 'ubs' meta-runner and 'install.sh' perform several external network requests to fetch dependencies and updates. It downloads language-specific scanner modules and AST helpers from the vendor's repository and fetches third-party binaries like 'ast-grep', 'ripgrep', and 'jq' from their official GitHub releases. The tool includes a robust checksum verification system using hardcoded SHA-256 digests to ensure the integrity of these downloads.
[COMMAND_EXECUTION]: The skill performs extensive local command execution to function. The 'install.sh' script creates directories, modifies shell configuration files (e.g., .bashrc, .zshrc), and installs git hooks. The 'ubs' script dispatches language modules that execute shell commands to run various scanners, and the installer may use 'sudo' for system-wide deployments.
[PROMPT_INJECTION]: The 'AGENTS.md' file includes authoritative instructions for AI agents working in the codebase, such as 'Rule 0
The Fundamental Override Prerogative,' which uses language intended to override an agent's default protocols ('YOU MUST LISTEN TO ME. I AM IN CHARGE, NOT YOU'). This is contextually presented as a project-specific guideline for agent-assisted development.
[SAFE]: The 'test-suite/' directory contains numerous files with intentionally buggy and insecure patterns, including SQL injection and unsafe deserialization ('pickle.loads'). These are clearly marked as test fixtures for the scanner and do not represent vulnerabilities in the tool's operational code.

ubs