#knowledge-graph #memory-system #vector-store #second-brain

Memory Pro System

OpenClaw Gateway plugin — enhanced AI memory with vector store, document-level MSA, knowledge graph, collision engine, and executable skills.

FluffyAIcode@fluffyaicode

Install

openclaw plugins install clawhub:openclaw-plugin-memory-pro

openclaw-plugin-memory-pro

OpenClaw Gateway plugin that adds an enhanced AI memory system — vector store (Memora), document-level MSA, knowledge graph, 7-strategy collision engine, and executable skills with closed-loop evolution.

Installation

bash

openclaw plugins install openclaw-plugin-memory-pro

Configuration

Add to your openclaw.json:

json

{
  "plugins": {
    "entries": {
      "memory-pro": {
        "enabled": true,
        "config": {
          "pythonPath": "python3",
          "memoryServerPort": 18790,
          "workspacePath": "/path/to/memory-pro-workspace",
          "autoStart": true,
          "contextInjection": true,
          "contextMaxTokens": 2000
        }
      }
    }
  }
}

Config Options

Option	Type	Default	Description
`pythonPath`	string	`"python3"`	Path to the Python 3 interpreter
`memoryServerPort`	number	`18790`	Port the memory server listens on
`workspacePath`	string	`""`	Path to the memory-pro workspace (auto-detected if empty)
`autoStart`	boolean	`true`	Auto-start the memory server on gateway load
`contextInjection`	boolean	`true`	Inject recalled memory into agent prompts
`contextMaxTokens`	number	`2000`	Max tokens for injected memory context

Agent Tools

The plugin registers 5 agent tools:

Tool	Description
`memory_remember`	Store information with automatic routing to Memora, MSA, and Chronos
`memory_recall`	Three-layer assembled retrieval with token budget control
`memory_deep_recall`	LLM-powered multi-hop deep recall
`memory_collide`	7-strategy attention-aware inspiration collision
`memory_skills`	List active skills with utility stats and executable prompts

Context Injection

When contextInjection is enabled, the plugin hooks into before_prompt_build to automatically recall relevant memories and prepend them to the agent's context.

HTTP Routes

Route	Auth	Description
`GET /memory-pro/health`	gateway	Plugin + memory server health
`GET /memory-pro/status`	gateway	Full memory system status

Prerequisites

Python 3.9+
The memory-pro workspace cloned and set up
An LLM API key (OpenRouter preferred, xAI fallback)

License

MIT