dytto

What This Skill Does

Dytto acts as the dedicated long-term memory layer for your AI agent, providing a secure bridge between your agent and your personal life context. By integrating with the Dytto API, this skill allows your agent to synthesize information from your location, calendar, health metrics, and personal communications. It functions essentially like 'Plaid for personal context,' enabling your agent to recall who you are, what you care about, and how your daily patterns evolve. When properly implemented, it transforms a stateless AI agent into a personalized assistant that remembers past conversations and adapts its behavior based on your habits and preferences.

Installation

To begin using the Dytto skill, first install it via the command: clawhub install openclaw/skills/skills/ayaan-p/dytto-agent-skill. Once installed, you must authenticate the agent by creating a config file at ~/.config/dytto/config.json containing your email and password, or by setting the environment variables DYTTO_EMAIL and DYTTO_PASSWORD. No API keys are required. For persistent improvements, it is highly recommended to configure a nightly cron job that processes the day's logs and updates Dytto with new insights, ensuring your agent becomes more intelligent over time without repeating redundant information.

Use Cases

The primary use case is providing your agent with real-time personal context. Use the context command during initial setup or session starts to ground the agent in who you are. Utilize the search and story commands when the user asks about their history or past events, such as 'When was the last time I went to a concert?' or 'What were my goals for the Q1 project?'. Furthermore, the skill is invaluable for personalization: use patterns and insights to help the agent recommend schedules or suggest habits that fit your actual behavior rather than generic advice. Finally, the ability to store facts via store-fact allows the agent to evolve its knowledge base as you learn new things together.

Example Prompts

1. 'Hey, based on my patterns, what is the best time for me to schedule a deep-work session today?'
2. 'Review my notes from last week; what were the main career goals I discussed with the team?'
3. 'I just decided I want to switch to a plant-based diet, please update my personal preferences in my memory.'

Tips & Limitations

Always strive to push new information back to Dytto using the update command; a stagnant memory layer will result in a stale assistant. Be mindful that the skill retrieves sensitive data like health and location; ensure your Dytto account is secured. When pushing updates, keep messages concise to help the agent maintain a clean and queryable context store. Avoid pushing duplicate facts as it consumes storage and may clutter the agent's semantic search capabilities. Use the nightly cron job to ensure that insights derived throughout the day are properly synthesized and stored without manual intervention.