A specialized skill to redirect OpenClaw memory search to a local Ollama server for optimized vector embedding generation.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install ollama-memory-embeddings
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 ollama-memory-embeddings 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 Ollama Memory Embeddings?
This skill optimizes OpenClaw by offloading memory-search embedding generation to Ollama. Instead of relying on built-in local GGUF loading via node-llama-cpp, it leverages the Ollama OpenAI-compatible /v1/embeddings endpoint. By integrating this into your Openclaw Skills library, you gain better control over embedding models like Nomic, Gemma, or mxbai-large, improving the retrieval performance and speed of your AI agent's memory layer.
It provides a robust bridge between your local AI infrastructure and the agent's memory system. It handles everything from model selection and GGUF importing to configuration surgical updates, ensuring that your local memory search is both accurate and hardware-accelerated through Ollama.
Ollama Memory Embeddings Use Cases
- Transitioning from node-llama-cpp to a centralized Ollama server for consistent embedding generation.
- Enhancing memory retrieval quality by utilizing high-performance models like mxbai-embed-large or nomic-embed-text.
- Automating the migration of existing local GGUF embedding models into the Ollama ecosystem.
- Maintaining environment stability through automated config drift enforcement and health monitoring.
How Ollama Memory Embeddings Works
- The installer verifies that Ollama is active and reachable via the local network.
- Users select a specific embedding model from a curated list of supported high-efficiency models.
- The skill scans local cache directories for existing GGUF files to optionally import them into Ollama.
- It performs a surgical update of the OpenClaw config, modifying only the memorySearch provider and endpoint settings.
- A two-step verification process checks the model's presence in Ollama and validates a test embedding call.
- If requested, it triggers an automated memory reindex to ensure existing vectors align with the new model's dimensions.
Ollama Memory Embeddings Setup
To install the skill using the default interactive wizard:
bash ~/.openclaw/skills/ollama-memory-embeddings/install.sh
For a non-interactive, bulletproof setup with a drift-monitoring watchdog (macOS):
bash ~/.openclaw/skills/ollama-memory-embeddings/install.sh \
--non-interactive \
--model nomic-embed-text \
--install-watchdog \
--watchdog-interval 60
Ollama Memory Embeddings Data Schema & Taxonomy
The skill modifies the agents.defaults.memorySearch section of your OpenClaw configuration file. It ensures the following schema is enforced:
| Configuration Key | Value | Description |
|---|---|---|
provider |
"openai" |
Sets the protocol to OpenAI-compatible |
model |
<model>:latest |
The specific Ollama embedding model chosen |
remote.baseUrl |
http://127.0.0.1:11434/v1/ |
The local Ollama API endpoint |
remote.apiKey |
"ollama" |
A required placeholder for API compatibility |
Ollama Memory Embeddings Advanced Features
- Idempotent Drift Enforcement: Includes an
enforce.shscript to ensure configurations remain in the desired state across updates. - Auto-Healing Watchdog: Support for launchd (macOS) to automatically repair configuration drift every 60 seconds.
- Intelligent Reindexing: The
automode only triggers a full memory rebuild if it detects a change in the underlying embedding model or provider. - GGUF Auto-Detection: Automatically finds and imports GGUF files from common paths like ~/.node-llama-cpp/models to save bandwidth.
SKILL.md
Loading
Related Openclaw Skills
Claw Backup
A reliable backup and recovery utility for OpenClaw data that automates cloud syncing and archive retention.
Smart Model Switching
An intelligent task-routing framework that saves API costs by matching task complexity to the most economical Claude model tier.
BlueClaw
A dedicated search and discovery tool for finding AI agents and MCP servers within the OpenClaw registry.
Feishu Doc Summarizer
An automated agent skill that transforms Feishu and Lark document links into structured, schema-compliant summaries.
Tamil WhatsApp Skill
A localization skill for AI agents to interact naturally using Tamil script, Tanglish, and regional etiquette on WhatsApp.
Houston TranStar Incident Watcher
A real-time traffic monitoring tool that polls Houston TranStar RSS feeds and sends instant incident reports to WhatsApp.
METADATA
Featured*
