Agent Customization

Your Agents, Your Rules

Every AI agent in FrankenCoder is fully customizable. Don't like how an agent behaves? Change it. Want to restrict what tools it can use? Toggle them off. All from inside the IDE, with changes taking effect immediately.

System Prompt Editing

Each agent has a system prompt that defines its personality, capabilities, and behavior. In FrankenCoder, you can edit these directly:

• Planner Agent: Adjust how it breaks down tasks
• Executor Agent: Change coding style preferences
• Reviewer Agent: Define what it checks for
• Security Agent: Customize security rules
• Research Agent: Modify research depth and sources
• All 13 agents: Fully customizable

Tool Management

Every agent has access to a set of tools. Sometimes you want to restrict what an agent can do:

• File operations: Read, write, delete permissions
• Terminal access: Command execution
• Browser control: Web navigation
• API calls: External service access
• Git operations: Version control actions

Just check or uncheck a box. The agent's capabilities update instantly.

Why This Matters

• Safety: Restrict dangerous operations when needed
• Focus: Limit agents to specific tasks
• Experimentation: Try different prompt strategies
• Specialization: Tune agents for your workflow
• Control: You're in charge, not the AI

On-the-Fly Changes

No restart required. No config files to edit. No redeployment. Just:

• Open agent settings
• Edit the prompt or toggle tools
• Save
• Agent immediately uses new settings

Use Cases

• Safer automation: Disable file deletion for overnight runs
• Code style: Make Executor follow your conventions
• Stricter reviews: Make Reviewer more thorough
• Research focus: Limit Research Agent to specific sources
• Teaching: Restrict tools for learning environments