codex-coding-plugin

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute specific host CLI tools for version verification and integration testing. These commands are localized to the development environment and are standard for the described purpose. Evidence includes usage of projectatlas --format json runtime-info, opencode --version, and claude mcp get projectatlas.
  • [SAFE]: The skill provides explicit safety guidance, instructing the agent to filter out unrelated environment variables from host debug output to prevent unintentional data exposure. It also promotes robust coding practices by requiring structured JSON parsing for configuration files.
  • [SAFE]: Surface analysis for data ingestion: The skill processes local configuration files (.json) using host-provided source paths and official vendor checks, minimizing the risk of processing malicious data. Ingestion points include .codex-plugin/plugin.json, .claude-plugin/plugin.json, and opencode/opencode.json. Boundary markers are established via official source verification logic, and capabilities are limited to local configuration and host CLI operations. Sanitization is enforced through the mandatory use of structured JSON parsers rather than string manipulation.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 05:35 PM
Security Audit — agent-trust-hub — codex-coding-plugin