codex-cli-bridge
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- COMMAND_EXECUTION (SAFE): The skill is architected to facilitate the use of the
codexCLI tool. It provides Python wrappers and documentation for executing shell-based coding tasks (e.g.,codex exec). This functionality is the primary intended purpose of the skill and is implemented through transparent command-line patterns. - EXTERNAL_DOWNLOADS (SAFE): The skill identifies
PyYAMLas a dependency in its documentation and scripts. This is a standard, reputable library for parsing YAML frontmatter. No suspicious or unversioned remote script downloads (e.g.,curl | bash) were detected. - INDIRECT_PROMPT_INJECTION (LOW): The bridge processes project metadata and instructions from
CLAUDE.mdto generate documentation for other tools. Ingestion points:CLAUDE.mdand project file structure. Boundary markers: Not explicitly enforced in the generated markdown. Capability inventory: Shell command execution via thecodexCLI. Sanitization: No explicit sanitization of input text before propagation toAGENTS.md. While this establishes a potential surface for instruction propagation, it is characteristic of documentation generators and is categorized as a low-risk concern. - PRIVILEGE_ESCALATION (SAFE): The included packaging script (
package.sh) useschmod +xto make the main Python script executable. This is standard practice for local development tools and does not constitute a security bypass.
Audit Metadata