agent-memory-store
Shared semantic memory store for AI agents. Store, search, and retrieve memories across agents with TTL decay. SQLite persistence — survives restarts.
Why use this skill?
Enhance your OpenClaw agents with persistent SQLite-backed semantic memory. Easily store, search, and manage data with TTL decay.
Install via CLI (Recommended)
clawhub install openclaw/skills/skills/kgnvsk/agent-memory-storeWhat This Skill Does
The agent-memory-store skill provides OpenClaw AI agents with persistent, long-term semantic memory capabilities. Unlike standard ephemeral agent state, this skill utilizes a SQLite-backed database located at /root/.openclaw/workspace/data/agent_memory.db, ensuring that stored information persists across system restarts. It acts as a central repository where agents can record key facts, user preferences, or task-specific metadata. The skill supports semantic search using Jaccard similarity, allowing agents to retrieve relevant information even when the exact phrasing differs. A core feature is the built-in Time-To-Live (TTL) decay, which allows for the automatic expiration of stale information, ensuring the memory store remains relevant and decluttered over time. By enabling cross-agent sharing, this skill facilitates sophisticated multi-agent workflows where multiple agents can contribute to and query a shared knowledge base.
Installation
To integrate this memory capability into your OpenClaw environment, use the command line interface to install the package from the official source repository. Execute the following command in your terminal:
clawhub install openclaw/skills/skills/kgnvsk/agent-memory-store
Once installed, ensure the memory service is running by executing python3 scripts/memory_store.py. This starts the lightweight HTTP server on port 8768, which will handle all incoming read and write requests from your agents.
Use Cases
- User Preference Tracking: Store user preferences like currency settings, coding styles, or recurring task parameters.
- Multi-Agent Coordination: Allow a 'researcher' agent to store findings and a 'writer' agent to retrieve them to synthesize a final report.
- Temporary Context Management: Store data with short TTLs to manage state during complex multi-step workflows that exceed standard prompt window limits.
Example Prompts
- "Store in the memory bank that the user prefers all responses to be formatted in markdown with a concise executive summary at the top."
- "Search through the agent memory store to find any previously noted preferences regarding payment methods for this project."
- "List the last 5 memories stored under the 'agent-beta' owner tag to verify the data sync."
Tips & Limitations
To maintain optimal performance, avoid storing large binary blobs in the memory store, as it is designed for textual semantic data. The current search implementation uses Jaccard similarity, which is efficient but may not capture complex vector-based semantic relationships as effectively as transformer-based embeddings. Always set appropriate TTL values to prevent the database from growing indefinitely, which could eventually impact retrieval performance. Since this skill interacts with a local SQLite file, ensure your OpenClaw workspace permissions allow for write access to the data directory to prevent errors during operation.
Metadata
Not sure this is the right skill?
Describe what you want to build — we'll match you to the best skill from 16,000+ options.
Find the right skillPaste this into your clawhub.json to enable this plugin.
{
"plugins": {
"official-kgnvsk-agent-memory-store": {
"enabled": true,
"auto_update": true
}
}
}Tags(AI)
Flags: file-write, file-read, network-access