Connects OpenClaw to AOMS (Always-On Memory Service) via HTTP, providing persistent tiered memory with semantic search, procedural learning, and episodic recall.

Features

🔍 Semantic search - Query memory with natural language
💾 Tiered storage - Episodic, semantic, procedural, working memory
⚡ Fast HTTP API - Local-first, sub-100ms queries
🔒 Optional auth - API key support via environment variables
📊 Weight tuning - Adjust memory importance dynamically

Installation

npm install openclaw-memory-aoms

Configuration

Add to your OpenClaw config (YAML or JSON):

YAML (`~/.openclaw/config.yml`)

plugins:
  - id: openclaw-memory-aoms
    config:
      baseUrl: http://localhost:9100
      timeoutMs: 10000
      apiKey: ${AOMS_API_KEY}  # Optional, reads from environment

JSON (`~/.openclaw/config.json`)

{
  "plugins": [
    {
      "id": "openclaw-memory-aoms",
      "config": {
        "baseUrl": "http://localhost:9100",
        "timeoutMs": 10000,
        "apiKey": "${AOMS_API_KEY}"
      }
    }
  ]
}

Configuration Options

Option	Type	Default	Description
`baseUrl`	string	`http://localhost:9100`	AOMS HTTP API endpoint
`timeoutMs`	number	`10000`	Request timeout (500-120000ms)
`apiKey`	string	-	Optional bearer token (supports `${ENV_VAR}` syntax)

Tools Provided

`memory_search`

Search memory entries across all tiers.

Parameters:

query (string, required) - Search query text
maxResults (number, optional) - Maximum results to return
minScore (number, optional) - Minimum relevance score threshold

Example:

// In your agent/skill code
const results = await tools.memory_search({
  query: "What did we decide about the API design?",
  maxResults: 5,
  minScore: 0.7
});

`memory_write`

Write a new memory entry to a specific tier.

Parameters:

tier (string, required) - Memory tier: episodic, semantic, procedural, working
text (string, required) - Memory content
metadata (object, optional) - Additional structured data

Example:

await tools.memory_write({
  tier: "episodic",
  text: "User prefers dark mode and compact UI layouts",
  metadata: {
    category: "preferences",
    confidence: 0.95
  }
});

`memory_weight`

Adjust the importance/weight of an existing memory entry.

Parameters:

id (string, required) - Memory entry ID
weight (number, required) - New weight value (typically -1 to 1)
tier (string, optional) - Tier hint for faster lookup

Example:

await tools.memory_weight({
  id: "mem_abc123",
  weight: 0.9  // Boost importance
});

Usage with AOMS

This plugin requires a running AOMS instance. Quick start:

# Clone AOMS
git clone https://github.com/yourusername/aoms.git
cd aoms

# Install dependencies
pip install -r requirements.txt

# Start AOMS HTTP server
python -m aoms.server --port 9100

See AOMS documentation for advanced setup (embeddings, vector stores, multi-tier config).

Development

# Clone this repo
git clone https://github.com/yourusername/openclaw-memory-aoms.git
cd openclaw-memory-aoms

# Install dependencies
npm install

# Build
npm run build

# Test locally (link into OpenClaw)
npm link
cd ~/.openclaw
npm link openclaw-memory-aoms

Architecture

OpenClaw Agent
      ↓
  Plugin Tools (memory_search, memory_write, memory_weight)
      ↓
  HTTP POST to AOMS API
      ↓
  AOMS Server (4-tier cortex, vector search, procedural learning)
      ↓
  ChromaDB / JSON storage

License

MIT

Contributing

PRs welcome! Please ensure:

TypeScript compiles without errors
No any types (use TypeBox schemas)
Follow existing code style
Update README if adding features

Made with 🧠 for autonomous AI agents

Keywords

FAQs

What is openclaw-memory-aoms?

Is openclaw-memory-aoms well maintained?

Package last updated on 28 Feb 2026

openclaw-memory-aoms