Syncing work context with Multi-AI workflows

Specific Hook Writing Instructions

Roast & Review Instructions

Unified Context Project Folder Setup

1. Open the terminal or command prompt on your computer.
2. Create a specific directory (folder) for your project using 'mkdir [nama_projek]'.
3. Navigate into the directory with the command 'cd [nama_projek]'.
4. Launch the CLI tools for Claude, Gemini, and Codex (or other custom agents) within the same directory.
5. Ensure each AI is opened in a separate terminal window but refers to the same folder path.
6. Prepare three main markdown files in the folder: 'gemini.md', 'claude.md', and 'agents.md' (or relevant .md files).
7. Ensure the initial content in each markdown file is synced to establish a consistent context baseline.

Executing a Simultaneous Multi-AI Workflow

1. Assign specific tasks to Claude (e.g., writing the Hook script) and instruct it to write the output directly to a specific '.md' file.
2. Assign a different task to Gemini (e.g., writing from a 'discovery' perspective) in another terminal window.
3. Use a Custom Agent (Codex) to observe and provide a real-time critique of the work produced by Claude and Gemini.
4. Monitor your folder to see new files being created automatically without needing to copy-paste between browsers.

'Script Session Closer' & Git Synchronization Procedure

1. Run the 'Session Closer' script (Custom Agent) once you're done working for the day.
2. Let the Agent collect all conversation logs and activities to generate a comprehensive summary.
3. Ensure the Agent updates the 'session summary' file for future reference.
4. Instruct the Agent to update all context files (Claude, Gemini, Agents) so the latest information is synchronized.
5. Perform a 'Git Commit' to save all project changes to a GitHub repo as an automated documentation system.