A sophisticated memory consolidation system inspired by Claude Code that organizes daily logs into persistent, categorized knowledge for AI agents.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install dreamclaw
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 dreamclaw 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 Auto Dream Skill?
Auto Dream is a powerful memory management framework designed to solve the problem of context loss in AI interactions. By implementing a structured architecture of logs and categorized memories, it allows agents to maintain a continuous understanding of users, projects, and feedback. This skill is a core component for anyone building advanced workflows using Openclaw Skills, as it bridges the gap between transient chat sessions and long-term knowledge retention.
The system operates on a dual-phase lifecycle: capturing information in real-time and periodically 'dreaming' to consolidate those insights. Unlike standard memory buffers, Auto Dream uses AI-driven distillation to prune outdated information and index relevant data, ensuring the agent's context remains lean, accurate, and highly relevant to the task at hand.
Auto Dream Skill Use Cases
- Tracking long-term user preferences and recurring project goals.
- Recording corrective feedback to prevent the AI from repeating previous mistakes.
- Maintaining technical specifications and architecture decisions for complex software projects.
- Indexing external references such as API endpoints, local database paths, and documentation pointers.
How Auto Dream Skill Works
- Append Phase: During a conversation, specific insights are appended to daily log files in a timestamped format.
- Orient Phase: When the dream command is triggered, the AI reviews the existing memory directory and the MEMORY.md index.
- Gather Phase: The system scans recent logs and session history to identify new information or contradictions with existing data.
- Consolidate Phase: The AI analyzes which entries deserve permanent storage, updates existing files, or deletes obsolete memories.
- Prune & Index Phase: The system updates the master MEMORY.md index to provide a high-level overview of all stored knowledge.
Auto Dream Skill Setup
To integrate this memory system into your environment, ensure the script is accessible in your local directory. You can interact with the system via the command line or by prompting the AI directly.
# Navigate to the skill directory
cd ~/.openclaw/skills/auto-dream/scripts
# Add a memory manually via CLI
python3 auto_dream.py append "User prefers Python 3.12 for all projects" --type project
# List all currently stored memories
python3 auto_dream.py list
Auto Dream Skill Data Schema & Taxonomy
The system organizes data within the .auto-dream/ directory using the following taxonomy:
| Path | Purpose | Format |
|---|---|---|
logs/YYYY/MM/ |
Daily append-only event logs | Markdown |
memories/user/ |
User profiles and preferences | YAML + Markdown |
memories/feedback/ |
Corrective guidance and instructions | YAML + Markdown |
memories/project/ |
Project-specific context and decisions | YAML + Markdown |
memories/reference/ |
External pointers and system data | YAML + Markdown |
MEMORY.md |
Root index of all consolidated memories | Markdown Index |
Auto Dream Skill Advanced Features
- Manual Dream Triggering: Users can initiate the consolidation process by simply saying 'dream' or 'organize memory' to the agent.
- Frontmatter Metadata: Every memory file includes YAML frontmatter (name, type, description, mtime) for structured data parsing.
- Conflict Detection: The AI-led distillation process identifies and resolves contradictory information between old memories and new logs.
- Automated Indexing: The system maintains a strict 200-line limit for the primary index file to ensure fast context loading for Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
AlphaEar Signal Tracker
A professional AI agent skill designed to monitor, assess, and update the evolution of financial investment signals based on real-time market data.
Luxury Travel Curator
A high-end AI concierge for designing bespoke luxury travel itineraries and exclusive global experiences.
Easy Travel with Child
An AI-powered family travel assistant that finds kid-friendly hotels and matches activities to your child's age and personality.
Smart Hotel Search
A powerful AI agent skill that combines social media discovery with transactional booking to handle non-standard hotel requests.
AlphaEar Reporter
A specialized skill for AI agents to plan, write, and edit professional financial reports with integrated chart configurations.
Multi-City Planner
An advanced travel optimization tool that automates complex multi-destination logistics and generates professional HTML itinerary reports.
Featured*
