Codex /goal — autonomous looping explained

Codex /goal — autonomous looping explained

Introduction to Cortex's /goal Feature

Enabling the /goal Feature

  • To activate the built-in /goal feature in Cortex, navigate to your computer's root directory and locate the .cortex folder. Open config.toml and add [features] goals = true. This will enable access to the /goals command upon booting up Cortex, along with a warning about under-development features.

Functionality of /goal

  • The /goal command allows users to set long-term objectives for their coding agents, enabling them to work on tasks over multiple sessions until completion. This contrasts with typical coding sessions where goals are often short-lived.

Importance of Task Lists and Validation Strategies

  • Integrating a comprehensive task list is crucial as it provides structure for the coding agent while pursuing its goal. Additionally, implementing a validation and testing strategy ensures that micro-tasks are validated before progressing further in achieving overarching goals. This structured approach enhances productivity and accuracy in coding efforts.

Learning Opportunities

  • For those interested in deepening their understanding of long-running coding agents and agentic orchestration, full courses and custom tools are available through the host's school community, with links provided in their bio for easy access.
Video description

The new Codex feature /goal allows for autonomous looping on high aiming project goals. #Codex #ChatGPT #OpenAI #ClaudeCode #Vibecoding