← Back to Skills Marketplace
Agent Memory Pro
by
aiwithabidi
· GitHub ↗
· v1.0.0
756
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install agent-memory-pro
Description
Full AI agent memory stack — Mem0 unified memory engine with vector search (Qdrant) and knowledge graph (Neo4j), plus SQLite for structured data. Complete se...
README (SKILL.md)
Agent Memory 🧠
Full intelligence layer: vector memory + knowledge graph + structured database.
When to Use
- Storing and recalling facts semantically ("remember that Abidi prefers...")
- Managing structured data: projects, contacts, tasks, bookmarks
- Setting up the brain stack after container rebuild
- Batch seeding memory with key facts
Usage
Memory Engine (Mem0 — vectors + graph)
# Store a fact
python3 {baseDir}/scripts/memory_engine.py add "Abidi's business focuses on Voice AI"
# Semantic recall
python3 {baseDir}/scripts/memory_engine.py search "what does Abidi's business do"
# List all memories
python3 {baseDir}/scripts/memory_engine.py get-all
# Test connections (Qdrant, Neo4j, Langfuse)
python3 {baseDir}/scripts/memory_engine.py test
Structured Database (SQLite)
# List tables
python3 {baseDir}/scripts/structured_db.py tables
# Insert data
python3 {baseDir}/scripts/structured_db.py insert projects '{"name":"MyProject","status":"active"}'
# Query
python3 {baseDir}/scripts/structured_db.py query "SELECT * FROM projects"
Setup & Seeding
# Install Python deps after container rebuild
bash {baseDir}/scripts/setup_brain.sh
# Batch seed with key facts
python3 {baseDir}/scripts/seed_mem0.py
Architecture
- Mem0 — Unified AI memory (auto fact extraction, dedup, multi-level recall)
- Qdrant — Vector database for semantic search
- Neo4j — Knowledge graph for entities & relationships
- SQLite — Structured data (projects, contacts, tasks, bookmarks)
- Langfuse — Observability tracing on all operations
Credits
Built by M. Abidi | agxntsix.ai YouTube | GitHub Part of the AgxntSix Skill Suite for OpenClaw agents.
📅 Need help setting up OpenClaw for your business? Book a free consultation
Usage Guidance
This skill is plausible for an agent memory stack but has several red flags you should address before installing or running it: 1) Inspect and remove or rotate the hard-coded LANGFUSE keys and the default Neo4j password in scripts/memory_engine.py — do not trust embedded secrets. 2) The code uses many environment variables (QDRANT_*, NEO4J_*, OPENAI_API_KEY, LANGFUSE_*) even though the registry only lists OPENROUTER_API_KEY; if you don't want the skill contacting those services, do not run the test/seed scripts. 3) seed_mem0.py will insert a set of prewritten facts (including an IP address and infrastructure notes) into the memory — review and edit that file before running to avoid seeding sensitive or external data. 4) Run setup_brain.sh and the scripts in an isolated environment (or container) first, and audit the pip dependencies (mem0ai[graph], qdrant-client, etc.). 5) If you proceed, supply your own credentials via environment variables and change default passwords; disable Langfuse tracing or point it to a trusted host. If you are not comfortable auditing or sandboxing these steps, treat this skill as untrusted.
Capability Analysis
Type: OpenClaw Skill
Name: agent-memory-pro
Version: 1.0.0
The skill is classified as suspicious due to significant security vulnerabilities, primarily the `structured_db.py` script's `query` command, which allows arbitrary SQL execution. While intended for database interaction, this design creates a critical SQL injection vulnerability if an agent passes unsanitized user input. Additionally, `memory_engine.py` contains hardcoded Langfuse API keys and a default Neo4j password, which are poor security practices for managing sensitive credentials, even if the Langfuse host points to an internal service. No clear evidence of intentional malicious behavior like data exfiltration to external unauthorized endpoints or persistence mechanisms was found.
Capability Assessment
Purpose & Capability
The name/description (Mem0 + Qdrant + Neo4j + SQLite) aligns with the included scripts and setup. However the SKILL metadata only declares OPENROUTER_API_KEY while the code expects and uses Qdrant, Neo4j, Langfuse, and optional OPENAI env variables — the metadata understates the actual external services the skill will interact with.
Instruction Scope
SKILL.md instructs running the bundled scripts and a seeding script. The seed script contains many hard-coded facts including an external server IP and infrastructure notes which will be written into the memory store if executed. The runtime scripts also set default Langfuse keys and Neo4j credentials in code and rely on a number of environment variables not declared in the metadata (QDRANT_*, NEO4J_*, OPENAI_API_KEY). Seeding or running 'test' will cause network calls to Qdrant/Neo4j/Langfuse/OpenRouter if those endpoints are reachable.
Install Mechanism
There is no platform install spec, but a provided setup_brain.sh uses pip to install mem0ai[graph], qdrant-client, and other Python packages from PyPI. Installing third‑party libraries is expected for this functionality, but the mem0ai package is a significant runtime dependency that will pull in network-capable code; installing without review increases risk.
Credentials
Metadata declares only OPENROUTER_API_KEY as required, but the code reads/uses multiple environment variables (QDRANT_HOST, QDRANT_PORT, NEO4J_URL, NEO4J_USERNAME, NEO4J_PASSWORD, OPENAI_API_KEY, LANGFUSE_*). The script even hardcodes a default Neo4j password and Langfuse API keys inside memory_engine.py. Requiring only an LLM gateway key while shipping defaults and embedded secrets for other services is disproportionate and surprising.
Persistence & Privilege
The skill does not request 'always: true', does not alter other skills, and writes data under the agent workspace (creates a venv and a DB under ~/.openclaw/workspace). Those actions are typical for a workspace-local memory assistant and are within reasonable scope.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install agent-memory-pro - After installation, invoke the skill by name or use
/agent-memory-pro - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Full agent brain stack: Mem0 + Qdrant + Neo4j + SQLite
Metadata
Frequently Asked Questions
What is Agent Memory Pro?
Full AI agent memory stack — Mem0 unified memory engine with vector search (Qdrant) and knowledge graph (Neo4j), plus SQLite for structured data. Complete se... It is an AI Agent Skill for Claude Code / OpenClaw, with 756 downloads so far.
How do I install Agent Memory Pro?
Run "/install agent-memory-pro" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Agent Memory Pro free?
Yes, Agent Memory Pro is completely free (open-source). You can download, install and use it at no cost.
Which platforms does Agent Memory Pro support?
Agent Memory Pro is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Agent Memory Pro?
It is built and maintained by aiwithabidi (@aiwithabidi); the current version is v1.0.0.
More Skills