📋 Define Your Goal
Tell the wizard what you want to accomplish on your system
What do you want to accomplish?
Be specific about file paths, services, and expected outcomes. The more detail you provide, the better the script will be.
💡 Tips for Better Results
- Be specific: Include file paths, service names, and exact commands when possible
- State the output: Describe where results should be saved or displayed
- Mention constraints: Note any safety requirements (e.g., "don't delete anything", "dry-run first")
- One goal at a time: Complex multi-step tasks work best as a single coherent goal
- Have a script already? Switch to Paste / Edit mode in the Pipeline tab to use your own script
⚙️ Pipeline
Generate, review, and execute your shell script
The AI reads your goal and writes a shell script (code/script.sh) to accomplish it.
Or, you can paste/edit your own script below and skip generation.
Paste or write your shell script below. Click Save Script to save it and proceed to Run & Fix.
Executes the generated script and automatically fixes any errors until it succeeds.
Results are logged to code/fix_log.md.
🚀 Run Entire Pipeline
Saves your goal, generates the script, then runs and fixes it — all in one click.
📊 Results
Review the generated script and execution output
Generated Shell Script
No script generated yet
Go to the Pipeline tab and run the pipeline to generate a shell script.
Execution & Fix Log
No execution log yet
Run the pipeline to see execution results and any auto-fix attempts.
Original Goal
No goal saved yet
Go to the Goal tab and describe what you want to accomplish.
💰 Usage
Track API token usage and estimated costs for this session
📊 Session Totals
🤖 Per-Model Breakdown
No usage data yet
Run the pipeline to see token usage and cost estimates here.
🔗 Task Session Usage
Usage from individual pipeline task sessions
No task sessions yet
Task session usage will appear here after running pipeline steps.
💡 Usage Tips
- Costs are estimates: Based on published model pricing; actual billing may vary
- Prompt tokens: Tokens sent to the model (your goal, context, instructions)
- Completion tokens: Tokens generated by the model (scripts, fixes, responses)
- Reduce costs: Use a fast/cheap model for the "Fast Model" setting and reserve expensive models for script generation
🔧 Settings
Configure AI models used for script generation and execution
🤖 Model Selection
Choose which AI models to use for generating and fixing scripts. Both models are required — the server does not provide defaults. Models are loaded from your configured API providers.
📋 Current Configuration
💡 Model Tips
- Smart Model: Used for the main script generation step — pick a capable model like GPT-4o or Claude Sonnet
- Fast Model: Used for parsing and auto-fix iterations — a smaller, faster model works well here
- Both models are required: The server does not provide defaults — you must select models before running the pipeline
- Image Model: Automatically set to the same as the Smart Model
- No models showing? Make sure you have at least one API provider configured with a valid API key
- Settings persist: Your model choices are saved in your browser and reused across sessions