VectorClaw is a security-hardened MyVector MySQL skill for OpenClaw that gives your AI agent persistent, intelligent memory. v4.0.0 introduces three powerful memory systems — HindSight, HoloGraphic, and Hancho — that work together to give your agent reasoning, reflection, and multi-dimensional recall.
What’s New in v4.0.0
HindSight — Post-Conversation Consolidation
HindSight gives your agent the ability to learn from conversations after they happen. It analyzes sentiment trends across recent interactions, identifies frequently discussed topics, discovers new subjects not yet stored, and promotes or demotes memories based on importance. Think of it as your agent reviewing its notes after a meeting and deciding what’s worth keeping.
HoloGraphic — Multi-Dimensional Memory Tagging
HoloGraphic tags every stored memory across multiple dimensions: emotion (positive/negative/complex), context (work/personal/health/tech/social), urgency (immediate/ongoing/historical), and people involved. This means your agent can answer questions like “How has Ev been feeling lately?” or “What health topics came up this month?” — not just keyword search.
Hancho — Memory That Reasons
Hancho goes beyond storage. It connects facts to derive new insights using 7 reasoning rules: medication side effects, health chains, tech infrastructure mapping, creative passion patterns, relationship depth indicators, interest-to-skill growth, and emotional coping patterns. When it finds a connection, it stores the derived insight as a new first-class memory with full lineage back to the source facts.
Hancho’s reasoning approach is inspired by honcho.dev — memory that doesn’t just store and retrieve, but actually thinks about what it knows.
How It Compares
| Feature | VectorClaw v4 | Mem0 | Supermemory |
|---|---|---|---|
| Storage | Self-hosted MySQL (MyVector) — you own your data | Cloud-hosted | Cloud-hosted |
| Recall | Semantic search + SQL queries + dimensional tags | Auto-recall during conversations | Auto-recall during conversations |
| Reasoning | ✅ Hancho knowledge graph with 7 derivation rules | ❌ Retrieval only | ❌ Retrieval only |
| Post-conversation analysis | ✅ HindSight consolidation every 6 hours | ❌ | ❌ |
| Multi-dimensional tags | ✅ HoloGraphic (emotion, context, urgency, people) | ❌ | ❌ |
| Sentiment tracking | ✅ Per-interaction + trend analysis | Basic | Basic |
| Engagement patterns | ✅ Time, day, topic, channel, session analysis | ❌ | ❌ |
| Self-hosted | ✅ Full control, no API limits | ❌ Cloud dependency | ❌ Cloud dependency |
| Cost | Free (self-hosted) | Free tier + paid | Free tier + paid |
| Data portability | ✅ Standard MySQL — export anytime | Vendor lock-in | Vendor lock-in |
The Memory Stack
VectorClaw v4 works best as part of a complete memory architecture:
- MEMORY.md — Always-in-context narrative (relationships, lessons, emotional context)
- VectorClaw (MyVector) — Structured profiles, interactions, dimensional tags, Hancho reasoning insights
- Mem0 — Auto-captured conversational facts with temporal decay (complementary)
- ChromaDB — Semantic search across workspace files (skills, projects, logs)
- HindSight + HoloGraphic + Hancho — Automated consolidation every 6 hours
Use VectorClaw alongside Mem0 for defense in depth — if one system misses a fact, the other catches it. The self-hosted nature means your most sensitive data never leaves your server.
Security
VectorClaw is built security-first. All SQL routes through a hardened wrapper enforcing: single-statement execution only, DDL blocking, table allowlists, comment injection prevention, hex-encoding detection, proper string escaping, least-privilege database users (root/admin rejected), and fail-closed authentication. 20/20 on independent security audit.
Requirements
- Docker (for MyVector container)
- OpenClaw agent
- ~500MB RAM for the MyVector container
Get It
VectorClaw is available as a free OpenClaw skill on ClawHub:
🔗 ClawHub: https://clawhub.ai/paradoxfuzzle/custom-mysql
Install with: openclaw skills install custom-mysql
🐼 Remember everything. Reason about it. Grow from it.