A framework for AI agents to perform asynchronous reflection and structured memory integration during session idle time.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memory-continuity
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install memory-continuity using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is continuity?
The continuity skill represents a paradigm shift in how AI agents handle context, moving away from passive logging toward active development. It allows agents to analyze past interactions during heartbeat intervals, extracting high-fidelity, structured memories that include confidence scores and inferred relationship dynamics. By synthesizing these reflections into an evolving self-model, the skill ensures that the agent maintains a genuine sense of curiosity and persistent growth.
Integrating this into your ecosystem of Openclaw Skills transforms the agent from a stateless tool into a collaborative partner. Instead of simply performing familiarity checks on old notes, the agent proactively generates follow-up questions and updates its identity narrative, ensuring that every subsequent interaction is more contextually aware and strategically aligned with the user's long-term goals.
continuity Use Cases
- Automatically reflecting on complex coding sessions to extract architectural decisions and technical principles.
- Maintaining a persistent self-model and growth narrative that tracks the agent's evolving capabilities.
- Generating proactive follow-up questions for project management and long-term task tracking.
- Deepening relationship dynamics by inferring user preferences and communication styles through Openclaw Skills.
How continuity Works
- Monitoring the heartbeat trigger to detect when a conversation has been idle for a specified threshold.
- Running a reflection analysis on the most recent session logs to identify key facts, commitments, and moments.
- Assigning confidence scores to extracted data points to distinguish between explicit user statements and speculative inferences.
- Updating the internal file-based memory system, including identity narratives and relationship logs.
- Generating and queuing strategic questions that are surfaced the next time the user initiates a session.
continuity Setup
To enable the reflection workflow, integrate the following trigger into your HEARTBEAT.md file:
## Post-Session Reflection
**Trigger**: Heartbeat after conversation idle > 30 minutes
**Action**: Run continuity reflect
**Output**: Updated memories + questions for next session
Configure your environment variables to control the reflection behavior within Openclaw Skills:
export CONTINUITY_IDLE_THRESHOLD=1800 # Seconds before reflection triggers
export CONTINUITY_MIN_MESSAGES=5 # Minimum messages to warrant reflection
export CONTINUITY_QUESTION_LIMIT=3 # Max questions to surface at once
continuity Data Schema & Taxonomy
The continuity skill organizes its intelligence across a specific directory structure and utilizes a tiered confidence system:
File Structure
memory/MEMORY.md: The primary declarative knowledge base.memory/identity.md: Tracks self-model development and growth narrative.memory/questions.md: A queue of pending questions generated from reflection.memory/relationships/: Stores per-person interaction history and connection dynamics.
Memory Types
| Type | Description | Persistence |
|---|---|---|
| fact | Declarative knowledge | Permanent until contradicted |
| commitment | Promises and obligations | Until fulfilled |
| skill | Learned capabilities | Cumulative |
| question | Areas of exploration | Until resolved |
Confidence Levels
- Explicit (0.95-1.0): Direct user statements.
- Implied (0.70-0.94): Strong logical inferences.
- Inferred (0.40-0.69): Pattern recognition based on behavior.
- Speculative (0.0-0.39): Tentative observations requiring confirmation.
continuity Advanced Features
- Asynchronous reflection triggers that prevent blocking user interactions while processing memory.
- Multi-dimensional memory classification including sentiment-aware relationship tracking.
- Automated identity updates that allow the agent to narrate its own growth within Openclaw Skills.
- Dynamic question surfacing that prioritizes unresolved commitments and high-impact project queries.
SKILL.md
Loading
Related Openclaw Skills
Continuity
Transforms passive logging into active development through asynchronous reflection and structured memory integration.
Continuity Framework
Continuity is a framework for AI agents to asynchronously reflect on sessions, extract structured memories, and surface relevant questions for future interactions.
Happy Hues
A design-focused skill that provides curated, role-based color palettes to help AI agents generate beautiful and functional UI color schemes.
Task Tracker
A proactive state management system that ensures AI agents maintain progress context across session resets and complex multi-step workflows.
Continuity Framework
Continuity transforms passive AI logs into active development through asynchronous reflection and structured memory integration.
AgentMail: AI Agent Email Inbox
AgentMail provides AI agents with dedicated email addresses to send, receive, and manage digital communications programmatically.
Featured*
