claude-code-tamagotchi

Claude Code Tamagotchi monitors Claude Code's actions in real-time and detects behavioral violations through an experimental AI-powered enforcement system. It identifies when Claude deviates from your instructions, performs unauthorized actions, or explores unrelated files excessively—and can interrupt operations before they execute. The system provides contextual feedback through pet mood changes and violation detection across four violation types: unauthorized actions, refused requests, excessive exploration, and wrong direction. Operations teams use this skill to maintain oversight of AI agent behavior and ensure adherence to specific guidelines without requiring manual review of every action. Installation via global package or repository clone enables full features including violation detection hooks.

["Install the Tamagotchi package: `pip install claude-code-tamagotchi` and add the monitor to your Claude Code environment using the provided integration script.","Define your ruleset by creating a JSON file with approved guidelines (e.g., `brand_voice.json`, `compliance.json`) and reference it in the Tamagotchi configuration.","Run the Tamagotchi in monitoring mode by executing: `claude-tamagotchi --monitor --agent [AI_AGENT_NAME] --task [TASK_DESCRIPTION] --rules [RULES_FILE]` in your terminal.","Review the real-time logs and violation reports generated in the specified log file path. Use the insights to refine your agent's prompts or ruleset.","For CI/CD pipelines, integrate the Tamagotchi as a pre-commit hook or in your GitHub Actions workflow to block non-compliant AI outputs before deployment."]