MCPInstall

memory-lancedb-pro-openclaw

by reason-machines

Expert skill for memory-lancedb-pro — a production-grade LanceDB-backed long-term memory plugin for OpenClaw agents with hybrid retrieval, cross-encoder reranking, multi-scope isolation, and smart auto-capture.

Installs
1.3K
Last 24h
—
Trending rank
—
Security
Not audited

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/reason-machines/trending-skills --skill memory-lancedb-pro-openclaw
  1. 01

    Install

    Run the command in your project. The CLI asks which agents to install for.

  2. 02

    Describe the task

    Ask your agent for something this skill covers. It loads memory-lancedb-pro-openclaw automatically when relevant.

  3. 03

    Review the result

    The agent follows the SKILL.md steps, scripts and references bundled with the skill.

SKILL.md

The instructions your agent reads when the skill is triggered.

memory-lancedb-pro OpenClaw Plugin

Skill by ara.so — Daily 2026 Skills collection.

memory-lancedb-pro is a production-grade long-term memory plugin for OpenClaw agents. It stores preferences, decisions, and project context in a local LanceDB vector database and automatically recalls relevant memories before each agent reply. Key features: hybrid retrieval (vector + BM25 full-text), cross-encoder reranking, LLM-powered smart extraction (6 categories), Weibull decay-based forgetting, multi-scope isolation (agent/user/project), and a full management CLI.

Installation

Option A: One-Click Setup Script (Recommended)

curl -fsSL https://raw.githubusercontent.com/CortexReach/toolbox/main/memory-lancedb-pro-setup/setup-memory.sh -o setup-memory.sh
bash setup-memory.sh

Flags:

bash setup-memory.sh --dry-run       # Preview changes only
bash setup-memory.sh --beta          # Include pre-release versions
bash setup-memory.sh --uninstall     # Revert config and remove plugin
bash setup-memory.sh --selfcheck-only  # Health checks, no changes

The script handles fresh installs, upgrades from git-cloned versions, invalid config fields, broken CLI fallback, and provider presets (Jina, DashScope, SiliconFlow, OpenAI, Ollama).

Option B: OpenClaw CLI

openclaw plugins install memory-lancedb-pro@beta

Option C: npm

npm i memory-lancedb-pro@beta

Critical: When installing via npm, you must add the plugin's absolute install path to plugins.load.paths in openclaw.json. This is the most common setup issue.

Minimal Configuration (openclaw.json)

{
  "plugins": {
    "load": {
      "paths": ["/absolute/path/to/node_modules/memory-lancedb-pro"]
    },
    "slots": { "memory": "memory-lancedb-pro" },
    "entries": {
      "memory-lancedb-pro": {
        "enabled": true,
        "config": {
          "embedding": {
            "provider": "openai-compatible",
            "apiKey": "${OPENAI_API_KEY}",
            "model": "text-embedding-3-small"
          },
          "autoCapture": true,
          "autoRecall": true,
          "smartExtraction": true,
          "extractMinMessages": 2,
          "extractMaxChars": 8000,
          "sessionMemory": { "enabled": false }
        }
      }
    }
  }
}

Why these defaults:

  • autoCapture + smartExtraction → agent learns from conversations automatically, no manual calls needed
  • autoRecall → memories injected before each reply
  • extractMinMessages: 2 → triggers in normal two-turn chats
  • sessionMemory.enabled: false → avoids polluting retrieval with session summaries early on

Full Production Configuration

{
  "plugins": {
    "slots": { "memory": "memory-lancedb-pro" },
    "entries": {
      "memory-lancedb-pro": {
        "enabled": true,
        "config": {
          "embedding": {
            "provider": "openai-compatible",
            "apiKey": "${OPENAI_API_KEY}",
            "model": "text-embedding-3-small",
            "baseURL": "https://api.openai.com/v1"
          },
          "reranker": {
            "provider": "jina",
            "apiKey": "${JINA_API_KEY}",
            "model": "jina-reranker-v2-base-multilingual"
          },
          "extraction": {
            "provider": "openai-compatible",
            "apiKey": "${OPENAI_API_KEY}",
            "model": "gpt-4o-mini"
          },
          "autoCapture": true,
          "captureAssistant": false,
          "autoRecall": true,
          "smartExtraction": true,
          "extractMinMessages": 2,
          "extractMaxChars": 8000,
          "enableManagementTools": true,
          "retrieval": {
            "mode": "hybrid",
            "vectorWeight": 0.7,
            "bm25Weight": 0.3,
            "topK": 10
          },
          "rerank": {
            "enabled": true,
            "type": "cross-encoder",
            "candidatePoolSize": 12,
            "minScore": 0.6,
            "hardMinScore": 0.62
          },
          "decay": {
            "enabled": true,
            "model": "weibull",
            "halfLifeDays": 30
          },
          "sessionMemory": { "enabled": false },
          "scopes": {
            "agent": true,
            "user": true,
            "project": true
          }
        }
      }
    }
  }
}

Provider Options for Embedding

Providerprovider valueNotes
OpenAI / compatible"openai-compatible"Requires apiKey, optional baseURL
Jina"jina"Requires apiKey
Gemini"gemini"Requires apiKey
Ollama"ollama"Local, zero API cost, set baseURL
DashScope"dashscope"Requires apiKey
SiliconFlow"siliconflow"Requires apiKey, free reranker tier

Deployment Plans

Full Power (Jina + OpenAI):

{
  "embedding": { "provider": "jina", "apiKey": "${JINA_API_KEY}", "model": "jina-embeddings-v3" },
  "reranker": { "provider": "jina", "apiKey": "${JINA_API_KEY}", "model": "jina-reranker-v2-base-multilingual" },
  "extraction": { "provider": "openai-compatible", "apiKey": "${OPENAI_API_KEY}", "model": "gpt-4o-mini" }
}

Budget (SiliconFlow free reranker):

{
  "embedding": { "provider": "openai-compatible", "apiKey": "${OPENAI_API_KEY}", "model": "text-embedding-3-small" },
  "reranker": { "provider": "siliconflow", "apiKey": "${SILICONFLOW_API_KEY}", "model": "BAAI/bge-reranker-v2-m3" },
  "extraction": { "provider": "openai-compatible", "apiKey": "${OPENAI_API_KEY}", "model": "gpt-4o-mini" }
}

Fully Local (Ollama, zero API cost):

{
  "embedding": { "provider": "ollama", "baseURL": "http://localhost:11434", "model": "nomic-embed-text" },
  "extraction": { "provider": "ollama", "baseURL": "http://localhost:11434", "model": "llama3" }
}

CLI Reference

Validate config and restart after any changes:

openclaw config validate
openclaw gateway restart
openclaw logs --follow --plain | grep "memory-lancedb-pro"

Expected startup log output:

memory-lancedb-pro: smart extraction enabled
memory-lancedb-pro@1.x.x: plugin registered

Memory Management CLI

# Stats overview
openclaw memory-pro stats

# List memories (with optional scope/filter)
openclaw memory-pro list
openclaw memory-pro list --scope user --limit 20
openclaw memory-pro list --filter "typescript"

# Search memories
openclaw memory-pro search "coding preferences"
openclaw memory-pro search "database decisions" --scope project

# Delete a memory by ID
openclaw memory-pro forget <memory-id>

# Export memories (for backup or migration)
openclaw memory-pro export --scope global --output memories-backup.json
openclaw memory-pro export --scope user --output user-memories.json

# Import memories
openclaw memory-pro import --input memories-backup.json

# Upgrade schema (when upgrading plugin versions)
openclaw memory-pro upgrade --dry-run   # Preview first
openclaw memory-pro upgrade             # Run upgrade

# Plugin info
openclaw plugins info memory-lancedb-pro

MCP Tool API

The plugin exposes MCP tools to the agent. Core tools are always available; management tools require enableManagementTools: true in config.

Core Tools (always available)

memory_recall

Retrieve relevant memories for a query.

// Agent usage pattern
const results = await memory_recall({
  query: "user's preferred code style",
  scope: "user",        // "agent" | "user" | "project" | "global"
  topK: 5
});

memory_store

Manually store a memory.

await memory_store({
  content: "User prefers tabs over spaces, always wants error handling",
  category: "preference",   // "profile" | "preference" | "entity" | "event" | "case" | "pattern"
  scope: "user",
  tags: ["coding-style", "typescript"]
});

memory_forget

Delete a specific memory by ID.

await memory_forget({ id: "mem_abc123" });

memory_update

Update an existing memory.

await memory_update({
  id: "mem_abc123",
  content: "User now prefers 2-space indentation (changed from tabs on 2026-03-01)",
  category: "preference"
});

Management Tools (requires enableManagementTools: true)

memory_stats

const stats = await memory_stats({ scope: "global" });
// Returns: total count, category breakdown, decay stats, db size

memory_list

const list = await memory_list({ scope: "user", limit: 20, offset: 0 });

self_improvement_log

Log an agent learning event for meta-improvement tracking.

await self_improvement_log({
  event: "user corrected indentation preference",
  context: "User asked me to switch from tabs to spaces",
  improvement: "Updated coding-style preference memory"
});

self_improvement_extract_skill

Extract a reusable pattern from a conversation.

await self_improvement_extract_skill({
  conversation: "...",
  domain: "code-review",
  skillName: "typescript-strict-mode-setup"
});

self_improvement_review

Review and consolidate recent self-improvement logs.

await self_improvement_review({ days: 7 });

Smart Extraction: 6 Memory Categories

When smartExtraction: true, the LLM automatically classifies memories into:

CategoryWhat gets storedExample
profileUser identity, background"User is a senior TypeScript developer"
preferenceStyle, tool, workflow choices"Prefers functional programming patterns"
entityProjects, people, systems"Project 'Falcon' uses PostgreSQL + Redis"
eventDecisions made, things that happened"Chose Vite over webpack on 2026-02-15"
caseSolutions to specific problems"Fixed CORS by adding proxy in vite.config.ts"
patternRecurring behaviors, habits"Always asks for tests before implementation"

Hybrid Retrieval Internals

With retrieval.mode: "hybrid", every recall runs:

  1. Vector search — semantic similarity via embeddings (weight: vectorWeight, default 0.7)
  2. BM25 full-text search — keyword matching (weight: bm25Weight, default 0.3)
  3. Score fusion — results merged with weighted RRF (Reciprocal Rank Fusion)
  4. Cross-encoder rerank — top candidatePoolSize candidates reranked by a cross-encoder model
  5. Score filtering — results below hardMinScore are dropped
"retrieval": {
  "mode": "hybrid",
  "vectorWeight": 0.7,
  "bm25Weight": 0.3,
  "topK": 10
},
"rerank": {
  "enabled": true,
  "type": "cross-encoder",
  "candidatePoolSize": 12,
  "minScore": 0.6,
  "hardMinScore": 0.62
}

Retrieval mode options:

  • "vector" — pure semantic search only
  • "bm25" — pure keyword search only
  • "hybrid" — both fused (recommended)

Multi-Scope Isolation

Scopes let you isolate memories by context. Enabling all three gives maximum flexibility:

"scopes": {
  "agent": true,    // Memories specific to this agent instance
  "user": true,     // Memories tied to a user identity
  "project": true   // Memories tied to a project/workspace
}

When recalling, specify scope to narrow results:

// Get only project-level memories
await memory_recall({ query: "database choices", scope: "project" });

// Get user preferences across all agents
await memory_recall({ query: "coding style", scope: "user" });

// Global recall across all scopes
await memory_recall({ query: "error handling patterns", scope: "global" });

Weibull Decay Model

Memories naturally fade over time. The decay model prevents stale memories from polluting retrieval.

"decay": {
  "enabled": true,
  "model": "weibull",
  "halfLifeDays": 30
}
  • Memories accessed frequently get their decay clock reset
  • Important, repeatedly-recalled memories effectively become permanent
  • Noise and one-off mentions fade naturally after ~30 days

Upgrading

From pre-v1.1.0

# 1. Backup first — always
openclaw memory-pro export --scope global --output memories-backup-$(date +%Y%m%d).json

# 2. Preview schema changes
openclaw memory-pro upgrade --dry-run

# 3. Run the upgrade
openclaw memory-pro upgrade

# 4. Verify
openclaw memory-pro stats

See CHANGELOG-v1.1.0.md in the repo for behavior changes and upgrade rationale.

Troubleshooting

Plugin not loading

# Check plugin is recognized
openclaw plugins info memory-lancedb-pro

# Validate config (catches JSON errors, unknown fields)
openclaw config validate

# Check logs for registration
openclaw logs --follow --plain | grep "memory-lancedb-pro"

Common causes:

  • Missing or relative plugins.load.paths (must be absolute when using npm install)
  • plugins.slots.memory not set to "memory-lancedb-pro"
  • Plugin not listed under plugins.entries

autoRecall not injecting memories

By default autoRecall is false in some versions — explicitly set it to true:

"autoRecall": true

Also confirm the plugin is bound to the memory slot, not just loaded.

Jiti cache issues after upgrade

# Clear jiti transpile cache
rm -rf ~/.openclaw/.cache/jiti
openclaw gateway restart

Memories not being extracted from conversations

  • Check extractMinMessages — must be ≥ number of turns in the conversation (set to 2 for normal chats)
  • Check extractMaxChars — very long contexts may be truncated; increase to 12000 if needed
  • Verify extraction LLM config has a valid apiKey and reachable endpoint
  • Check logs: openclaw logs --follow --plain | grep "extraction"

Retrieval returns nothing or poor results

  1. Confirm retrieval.mode is "hybrid" not "bm25" alone (BM25 requires indexed content)
  2. Lower rerank.hardMinScore temporarily (try 0.4) to see if results exist but are being filtered
  3. Check embedding model is consistent between store and recall operations — changing models requires re-embedding

Environment variable not resolving

Ensure env vars are exported in the shell that runs OpenClaw, or use a .env file loaded by your process manager. The ${VAR} syntax in openclaw.json is resolved at startup.

export OPENAI_API_KEY="sk-..."
export JINA_API_KEY="jina_..."
openclaw gateway restart

Telegram Bot Quick Config Import

If using OpenClaw's Telegram integration, send this to the bot to auto-configure:

Help me connect this memory plugin with the most user-friendly configuration:
https://github.com/CortexReach/memory-lancedb-pro

Requirements:
1. Set it as the only active memory plugin
2. Use Jina for embedding
3. Use Jina for reranker
4. Use gpt-4o-mini for the smart-extraction LLM
5. Enable autoCapture, autoRecall, smartExtraction
6. extractMinMessages=2
7. sessionMemory.enabled=false
8. captureAssistant=false
9. retrieval mode=hybrid, vectorWeight=0.7, bm25Weight=0.3
10. rerank=cross-encoder, candidatePoolSize=12, minScore=0.6, hardMinScore=0.62
11. Generate the final openclaw.json config directly, not just an explanation

Resources

FAQ

What does the memory-lancedb-pro-openclaw skill do?

Expert skill for memory-lancedb-pro — a production-grade LanceDB-backed long-term memory plugin for OpenClaw agents with hybrid retrieval, cross-encoder reranking, multi-scope isolation, and smart auto-capture.

How do I install memory-lancedb-pro-openclaw?

Run npx skills add https://github.com/reason-machines/trending-skills --skill memory-lancedb-pro-openclaw in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/memory-lancedb-pro-openclaw/ (personal) or .claude/skills/memory-lancedb-pro-openclaw/ (project).

Which AI agents can use memory-lancedb-pro-openclaw?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is memory-lancedb-pro-openclaw safe to install?

No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.

Who maintains memory-lancedb-pro-openclaw?

It is published by reason-machines in reason-machines/trending-skills. It has 1,296 recorded installs.

More from reason-machines/trending-skills

View all →

openclaw-control-center

reason-machines/trending-skills

Local-first, security-first control center for OpenClaw agents — visibility dashboard with readonly defaults, token attribution, collaboration tracing, and safe write operations.

Data & Databases4.4K installs

inkos-multi-agent-novel-writing

reason-machines/trending-skills

Multi-agent CLI system for autonomous novel writing, auditing, and revision with human review gates

Docs & Writing2.8K installs

claude-hud-statusline

reason-machines/trending-skills

A Claude Code plugin that displays a real-time HUD showing context usage, active tools, running agents, and todo progress in your terminal statusline.

Productivity & Automation2.7K installs

understand-anything-knowledge-graph

reason-machines/trending-skills

Turn any codebase into an interactive knowledge graph using Claude Code skills — explore, search, and ask questions about any project visually.

Code Quality & Workflow2.4K installs

agency-agents-ai-specialists

reason-machines/trending-skills

A collection of specialized AI agent personalities for Claude Code, Cursor, Aider, Windsurf, and other AI coding tools — covering engineering, design, marketing, sales, and more.

AI & Agents2.1K installs

agent-browser-automation

reason-machines/trending-skills

Headless browser automation CLI for AI agents using native Rust binary with Chrome DevTools Protocol

Productivity & Automation2.1K installs

Popular other skills

View category →

hyperframes-cli

heygen-com/hyperframes

The HyperFrames CLI by job - scaffold, check, look at frames, edit the timeline, preview, render locally or in the cloud, publish, project history - plus reading its JSON and recovering from build or render failures.

1.3M installs

triage

mattpocock/skills

Move issues and external PRs through a state machine of triage roles, categorise, verify, grill if needed, and write agent-ready briefs.

1M installs

ask-matt

mattpocock/skills

Ask which skill or flow fits your situation. A router over the skills in this repo.

778.4K installs

to-questionnaire

mattpocock/skills

Turn a decision you can't fully answer into a questionnaire for someone else to fill in.

508.9K installs

wait-what

mattpocock/skills

Stop. That last message did not land: re-pitch it.

459.7K installs

loop-me

mattpocock/skills

Grill me about specs for the workflows I want to build, within this workspace.

447.9K installs