Fractal Memory is an automated hierarchical compression system that distills raw agent experiences into structured patterns to prevent context overflow.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install fractal-memory
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 fractal-memory 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 Fractal Memory?
Fractal Memory solves the challenge of long-term agent persistence by mimicking the way human memory functions. Instead of allowing conversation logs to grow indefinitely, this skill organizes data into a hierarchy of daily, weekly, and monthly rollups. This ensures that the most relevant and distilled information is always available to the agent without hitting token limits.
By integrating this system into Openclaw Skills, developers can create agents that grow more capable over time. The system moves from raw experience to patterns, and from patterns to principles, maintaining the essence of interactions while discarding the noise of session bloat.
Fractal Memory Use Cases
- Managing long-term AI agent memory without requiring manual curation or intervention.
- Optimizing token usage by loading a hierarchical context instead of flat history files.
- Preventing session degradation in long-running autonomous workflows.
- Migrating fragmented daily logs into a structured, searchable memory architecture.
- Tracking evolving project trajectories and milestones over months of agent operation.
How Fractal Memory Works
- Interaction Capture: The agent writes real-time events and interactions directly to daily log files.
- Daily Rollup: At the end of each day, a script extracts key decisions and patterns into a weekly summary file.
- Weekly Compression: Every Sunday, the system distills weekly summaries into a monthly trajectory report.
- Core Distillation: On the last day of each month, the most significant lessons and themes are moved into the permanent MEMORY.md file.
- Context Retrieval: The agent loads context in a specific order—Today, This Week, This Month, then Core Memory—to optimize for recency and primacy.
Fractal Memory Setup
To initialize the fractal memory structure within your workspace, execute the following commands:
# Create the hierarchical directory structure
mkdir -p memory/diary/{2026/{daily,weekly,monthly},sticky-notes/{workflows,apis,commands,facts}}
# Initialize state tracking files
cp assets/rollup-state.json memory/
cp assets/heartbeat-state.json memory/
# Install core automation scripts
cp scripts/*.py ~/.openclaw/workspace/scripts/
chmod +x ~/.openclaw/workspace/scripts/*.py
After setup, configure your cron jobs to trigger the rollup scripts at 23:59 daily, weekly, and monthly to maintain the Openclaw Skills memory pipeline.
Fractal Memory Data Schema & Taxonomy
The system organizes data into a structured taxonomy to ensure efficient retrieval:
| Component | File Path | Purpose |
|---|---|---|
| Daily Logs | memory/diary/YYYY/daily/ |
Raw events and interaction logs. |
| Weekly Summaries | memory/diary/YYYY/weekly/ |
Compressed patterns and decisions. |
| Monthly Summaries | memory/diary/YYYY/monthly/ |
High-level milestones and trajectories. |
| Core Memory | MEMORY.md |
The essential index of agent identity and core facts. |
| Sticky Notes | memory/sticky-notes/ |
Categorized files for timeless technical facts (API keys, CLI commands). |
Fractal Memory Advanced Features
- Automated integrity verification using verify_memory_integrity.py to detect data tampering or corruption.
- Primacy and recency optimized context loading strategy for better LLM attention.
- Provenance tracking with metadata and timestamps for every compressed memory layer.
- Sticky-note categorization for managing non-chronological, timeless information.
- Programmatic memory injection via the append_to_daily.py utility for automated event logging.
SKILL.md
Loading
Related Openclaw Skills
HokiPoki
A CLI-based P2P tool for routing tasks to different AI models and sharing subscriptions securely within teams.
PayRam vs x402 Comparison Skill
A technical reference for comparing self-hosted PayRam infrastructure with the x402 HTTP payment protocol for AI agent commerce.
GitHub MCP Server
A comprehensive Model Context Protocol server that empowers AI agents to manage GitHub repositories, code files, issues, and pull requests autonomously.
Filesystem MCP Server
A secure, sandboxed Model Context Protocol server that enables AI agents to perform file and directory operations within explicitly allowed paths.
fdv.lol - Headless Memecoin Auto Trader
A CLI-based automated memecoin trader for Solana that leverages Agent Gary for autonomous AI-driven decision making.
SocialRails
A comprehensive social media management skill that enables post scheduling, AI caption generation, and analytics tracking via chat commands.
Featured*
