A stateless symbolic memory workflow that stores facts in PostgreSQL and activates semantics just-in-time for Ollama-powered LLM agents.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install symbolic-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 symbolic-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 Symbolic Memory?
Symbolic Memory is a sophisticated design pattern implemented as one of the Openclaw Skills to provide AI agents with a stateless yet durable knowledge substrate. Unlike traditional agent memory which often relies on local, ephemeral state, this skill utilizes a PostgreSQL backend to store canonical semantics and facts. By separating stored semantics from computed meaning, it ensures that agents can operate with long-term coherence and multi-agent consistency without bloating the LLM's context window.
The core philosophy behind this skill is to store knowledge as symbols and only activate their full meaning just-in-time (JIT). This budgeted approach to information retrieval allows for highly efficient processing within Ollama-based models, making it a critical tool for developers building complex, long-running agentic workflows using Openclaw Skills.
Symbolic Memory Use Cases
- Implementing long-term cross-session knowledge for AI agents without local state.
- Ensuring multi-agent consistency through a shared, versioned symbolic substrate.
- Managing massive knowledge bases by activating only relevant facts just-in-time.
- Reducing LLM context window noise by replacing raw data with symbolic references via Openclaw Skills.
How Symbolic Memory Works
- Canonical semantics and facts are ingested and stored within a structured PostgreSQL database.
- Knowledge points are assigned specific symbols which serve as lightweight references for the agent.
- Upon a specific trigger or query, the system performs JIT meaning activation based on a defined budget.
- Only the activated facts are injected into the Ollama LLM prompt, ensuring relevant and concise context processing within the Openclaw Skills framework.
Symbolic Memory Setup
To deploy this skill, ensure you have PostgreSQL and Python3 installed on your system. Configuration is managed via environment variables.
# Set your PostgreSQL Connection String
export PG_DSN="postgres://username:password@localhost:5432/memory_db"
# Configure Ollama Connectivity
export OLLAMA_HOST="http://localhost:11434"
export OLLAMA_MODEL="llama3"
# Ensure psql and python3 are in your PATH
which psql
which python3
Symbolic Memory Data Schema & Taxonomy
The skill organizes data using a relational SQL approach combined with semantic metadata.
| Component | Type | Description |
|---|---|---|
| SQL Facts | Relational Table | Stores the raw facts and ground-truth data points. |
| Canonical Semantics | Metadata | Defines the immutable meaning of specific symbols. |
| Symbols | Unique Identifiers | Short-hand references used by the agent to represent complex data. |
| Versioning | Metadata | Tracks changes to the symbolic substrate for multi-agent synchronization in Openclaw Skills. |
Symbolic Memory Advanced Features
- Budgeted Meaning Activation: Dynamically controls the amount of detail sent to the LLM to optimize token usage.
- Multi-Agent Synchronization: A shared PostgreSQL substrate allows multiple agents to access and contribute to a unified memory.
- Stateless Operations: Agents can be destroyed and recreated without losing access to their durable symbolic memory.
- Native Ollama Integration: Optimized for local LLM execution, ensuring data privacy and low latency for Openclaw Skills users.
SKILL.md
Loading
Related Openclaw Skills
KIS Trading
An AI coding agent skill for executing Korean stock trades and managing portfolios via the Korea Investment & Securities (KIS) Open API.
Test Audit Badge
A specialized demonstration skill designed to showcase and verify the audit badge functionality within the environment.
Clawcast EVM Wallet Manager
A comprehensive toolset for managing EVM wallets and executing blockchain transactions directly through an AI agent via Foundry cast.
Clawbrowser
A terminal-driven browser automation skill that allows AI agents to navigate the web, interact with elements, and manage persistent sessions via Playwright CLI.
LocalSquare: AI-Driven Local Business Advertising
LocalSquare allows AI agents to claim and pay for permanent digital bulletin board advertisements for local businesses across 43,000+ ZIP codes.
Agent Config Skill
A specialized workflow for intelligently managing and modifying the core configuration files that define an AI agent's behavior, personality, and operational procedures.
METADATA
Featured*
