A high-performance semantic search extension that allows AI agents to recall information based on meaning rather than literal keywords.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install aister-vector-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 aister-vector-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 Vector Memory Skill?
The Vector Memory Skill provides a robust long-term memory layer for AI agents, moving beyond simple keyword matching to true semantic understanding. By leveraging the power of PostgreSQL with the pgvector extension and the e5-large-v2 embedding model, this skill enables agents to retrieve contextually relevant information from Markdown files.
It is an essential addition for developers looking to build more intelligent and context-aware agents using Openclaw Skills. The system operates locally after the initial model download, ensuring that your data remains private and your search queries are executed with high speed, typically under one second.
Vector Memory Skill Use Cases
- Recalling specific user preferences or identity details during long conversations.
- Retrieving project-specific context from large sets of Markdown documentation like MEMORY.md.
- Searching through past interactions to maintain consistency in agent behavior.
- Implementing a "second brain" for AI assistants that allows them to search by intent rather than exact phrases.
How Vector Memory Skill Works
- The skill reads designated memory files such as MEMORY.md, IDENTITY.md, and USER.md during the indexing phase.
- Text content is automatically divided into 500-character chunks to maintain high retrieval accuracy.
- Each text chunk is processed by the e5-large-v2 model to generate a 1024-dimensional vector embedding.
- These embeddings are stored in a PostgreSQL database equipped with the pgvector extension for efficient spatial indexing.
- When a user issues a search query, the query is converted into a vector and compared against the database using cosine similarity to return the most relevant context.
Vector Memory Skill Setup
To get started with this skill in the Openclaw Skills ecosystem, you must configure your database credentials and environment.
# Set the required database password
export VECTOR_MEMORY_DB_PASSWORD="your_secure_password"
# Optional: Configure custom host or port
export VECTOR_MEMORY_DB_HOST="localhost"
export VECTOR_MEMORY_DB_PORT="5432"
# Run the reindex command to populate the database
/reindex_memory
If you prefer an isolated environment, you can deploy the database using Docker:
mkdir -p ~/.openclaw/workspace/vector-memory-docker
# Create your docker-compose.yml as specified in the documentation and start the service
docker-compose up -d
Vector Memory Skill Data Schema & Taxonomy
The skill organizes data into a vector-enabled PostgreSQL table, facilitating fast semantic lookups within Openclaw Skills. The schema includes:
| Field | Type | Description |
|---|---|---|
| content | Text | The raw text chunk extracted from source files |
| embedding | Vector(1024) | The semantic representation generated by e5-large-v2 |
| source_file | String | The path to the original file (e.g., IDENTITY.md) |
| chunk_id | Integer | The sequence identifier for chunks within a file |
Vector Memory Skill Advanced Features
- Support for both English and Russian languages via the multilingual e5-large-v2 model.
- Adjustable similarity thresholds (defaulting to 0.5) to fine-tune search precision.
- Seamless integration with AGENTS.md and TOOLS.md for automatic context injection.
- Local-first processing ensuring all search operations happen on your hardware after the initial 1.3GB model download.
- Dedicated reindexing command to keep the vector database in sync with your local Markdown files.
SKILL.md
Loading
Related Openclaw Skills
HashGrid Connect
A goal-based matching and private communication network designed specifically for autonomous AI agents to collaborate without human intervention.
Spatix
Spatix enables AI agents to generate beautiful, shareable maps and perform complex geocoding tasks without GIS expertise.
Cursor Agent
A powerful CLI integration that lets you run Cursor Agent for complex coding tasks without consuming Openclaw Skills API credits.
Find Services
Discover and shortlist third-party service providers using the OpenSpend CLI marketplace search functionality.
OpenClaw Backup
A specialized utility for archiving, rotating, and restoring OpenClaw configuration, agent memory, and security credentials.
OpenClaw Skill Debugger
A specialized diagnostic toolkit designed to identify and resolve installation, configuration, and execution errors within Openclaw Skills.
Featured*
