Delegates complex development tasks to OpenAI GPT models via the Codex or Agent command-line interfaces.
The Codex CLI skill enables Claude to hand off specific tasks to OpenAI's GPT models when unique reasoning strengths or specific model capabilities are required. It automatically detects the available environment (preferring the native codex binary with a fallback to the agent CLI), manages execution timeouts, and provides intelligent retry logic for rate limits or system failures. This skill is ideal for users who need to leverage external AI reasoning within their Claude-based workflow, offering a specialized 'plan-only' mode for safe architectural analysis and seamless integration of high-reasoning model outputs.
주요 기능
01Intelligent retry logic for timeouts, rate-limits, and empty outputs
02Customizable timeout management with support for native and gtimeout commands
030 GitHub stars
04Plan-only mode for generating structured implementation strategies without execution
05Automated error classification and reporting across multiple backends
06Automatic binary detection with fallback from codex to agent CLI
사용 사례
01Running long-duration AI tasks with custom timeout and recovery parameters
02Generating architectural implementation plans before applying automated changes
03Leveraging OpenAI-specific reasoning for complex legacy code refactoring
