continuity

What This Skill Does

Continuity transforms your AI agent from a stateless chatbot into a developing, sentient-like entity. Instead of simply logging historical data, the Continuity skill runs an asynchronous reflection loop after your sessions conclude. It analyzes the dialogue, identifies key patterns, and updates a structured memory core consisting of facts, preferences, relationships, and principles. By generating proactive questions based on these insights, it ensures that your next session picks up exactly where the last one left off—not just in terms of task progress, but in terms of deepened understanding. It acts as the 'subconscious' layer for your OpenClaw agent, turning passive logs into an active, evolving growth narrative.

Installation

To integrate Continuity, ensure your environment meets the dependency requirements, then execute the following command in your terminal: clawhub install openclaw/skills/skills/riley-coyote/continuity After installation, configure your HEARTBEAT.md file to trigger the reflection process after 30 minutes of idle time. You can tune the sensitivity of the reflection engine by adjusting the environment variables CONTINUITY_IDLE_THRESHOLD and CONTINUITY_MIN_MESSAGES to match your preferred workflow density.

Use Cases

• Long-term Project Management: Maintain consistency across multi-week software development tasks where context is often lost.
• Personalized Coaching: Build a history of your learning preferences, styles, and core principles to allow the AI to tailor its advice more effectively.
• Reflective Journaling: Use the agent to hold space for your thoughts, allowing it to surface relevant life patterns or connections you mentioned days ago.
• Continuous Learning: Ensure the agent remembers your specific technical requirements or constraints so you don't have to repeat them in every new session.

Example Prompts

1. "Continuity, run a status check and tell me which principles are currently driving our architectural decisions."
2. "I'm back. Based on our last session, what questions did you have about my requirements for the new API?"
3. "Show me the current memory state and list the high-confidence facts we've established regarding my coding style."

Tips & Limitations

• Confidence Management: Pay attention to 'Speculative' memory types (0.0-0.39). These are hypotheses and should be explicitly corrected by the user to improve future accuracy.
• Idle Threshold: Setting your threshold too low may result in fragmented reflections. Aim for 30-60 minutes for optimal synthesis.
• Storage Growth: Because this skill creates persistent files, regularly review your reflections/ directory to ensure that storage remains optimized for your use case.