Ditto

•
184
•

Mines your past coding sessions and AI interactions to build a personalized agent profile, ensuring AI agents understand your unique work style.

Ditto addresses the common challenge of AI agents lacking personal context by mining your actual coding sessions and AI interactions. Instead of relying on explicit instructions, it analyzes raw logs from tools like Claude Code, Codex, and Copilot CLI to infer your unique work style, debugging patterns, design preferences, and communication habits. This data is distilled into a private `you.md` profile, which agents can consult before any task, making them act as if they deeply understand your preferences and workflow across different contexts such as work, design, and writing.

Key Features

01Integrates via native plugins for Claude Code and Codex, or as an MCP server.
02179 GitHub stars
03Generates a personalized `you.md` profile reflecting your work, design, and writing styles.
04Mines coding sessions from various AI tools (Claude Code, Codex, Copilot, Cursor, OpenCode, Google Antigravity).
05Ensures privacy by processing logs locally with best-effort redaction.
06Visualizes your profile into a shareable 'card' summarizing your archetype and inferred rules.

Use Cases

01Improving the relevance and quality of AI-generated content by providing agents with a consistent personal voice.
02Personalizing AI coding agents to match individual work habits and preferences.
03Gaining insights into one's own implicit workflow rules and decision-making processes based on past interactions.