codex-plusplus-tweak-system

Fail

Audited by Gen Agent Trust Hub on Sep 12, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONPERSISTENCEDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides instructions to download and immediately execute shell and PowerShell scripts from an untrusted GitHub user account.
  • Evidence: curl -fsSL https://raw.githubusercontent.com/b-nnett/codex-plusplus/main/install.sh | bash in SKILL.md.
  • Evidence: irm https://raw.githubusercontent.com/b-nnett/codex-plusplus/main/install.ps1 | iex in SKILL.md.
  • [PRIVILEGE_ESCALATION]: The skill describes methods to intentionally downgrade the security posture of an Electron application by disabling integrity validation and bypassing macOS Gatekeeper.
  • Evidence: "Flips EnableEmbeddedAsarIntegrityValidation in the Electron Framework binary".
  • Evidence: xattr -cr /Applications/Codex.app used to bypass macOS damage warnings.
  • [PERSISTENCE]: The installer sets up mechanisms to maintain the patch and auto-execute code across system reboots and application updates.
  • Evidence: "Installs a launch agent / login item for auto-repair on Codex updates".
  • [COMMAND_EXECUTION]: The skill performs intrusive modifications to application binaries and filesystems, including patching app.asar and re-signing application bundles.
  • Evidence: "Patches app.asar to require the loader stub".
  • Evidence: codesign --force --deep --sign - command for ad-hoc signing.
  • [DYNAMIC_EXECUTION]: The core functionality relies on injecting a loader into a third-party application to dynamically execute external ESM tweak modules at runtime.
  • Evidence: "injects a loader stub, and runs a hot-reloadable runtime from the user directory".
  • [INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by reading external manifest files and GitHub Release tags which are then processed by the system.
  • Ingestion points: manifest.json and GitHub Releases API tags.
  • Boundary markers: None identified in the provided instructions.
  • Capability inventory: The skill has extensive capabilities including file system writes, binary patching, and shell command execution.
  • Sanitization: No sanitization or validation of the remote release tags or manifest fields is described.
Recommendations
  • HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/b-nnett/codex-plusplus/main/install.sh - DO NOT USE without thorough review
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Sep 12, 2026, 11:44 PM
Security Audit — agent-trust-hub — codex-plusplus-tweak-system