MidOS is a comprehensive knowledge middleware layer that provides AI agents with long-term memory, hybrid search, and multi-step planning capabilities.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install midos-mcp
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install midos-mcp using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is MidOS MCP?
MidOS MCP acts as a powerful knowledge operating system designed to function as the long-term brain for your AI agents. By integrating this skill from the Openclaw Skills collection, developers can give their agents access to a massive intelligence base comprising over 670,000 vectors and 46,000 knowledge chunks. It bridges the gap between ephemeral LLM context and persistent, structured knowledge management.
The system utilizes EUREKA synthesis to provide deep insights and high-coherence knowledge scoring, ensuring that your agents have access to high-quality data. Whether used for complex multi-agent orchestration or simple persistent memory, MidOS ensures that agents remain context-aware and data-driven across multiple sessions. Integrating Openclaw Skills like MidOS allows for a more robust and reliable AI developer experience.
MidOS MCP Use Cases
- Persistent cross-session memory to maintain user preferences, project context, and past decisions.
- Complex multi-step task planning where agents must track progress across context resets.
- High-precision hybrid search for technical documentation or research data using both BM25 and semantic vectors.
- Automated knowledge base maintenance through preflight checks and quality gate validation.
- Multi-agent orchestration utilizing shared memory spaces and stateful planning tools.
How MidOS MCP Works
- The agent initializes a session via the MCP JSON-RPC 2.0 protocol to establish a secure connection with the MidOS server.
- Agents utilize memory tools like mem_search or where_was_i to retrieve historical context and specific project data from previous sessions.
- For new incoming information, the knowledge_preflight tool checks for duplicates in milliseconds while the quality_gate evaluates content before storage.
- Planning tools enable the creation of structured task lists that the agent updates as it executes shell commands or file operations.
- All data is indexed using LanceDB and Gemini-based embeddings, enabling lightning-fast semantic and hybrid retrieval across the entire knowledge base.
MidOS MCP Setup
To start using this skill from Openclaw Skills, you can connect via the hosted API or set up a self-hosted instance.
Connect via MCP (JSON-RPC 2.0)
Initialize your session using a standard request:
curl -X POST https://midos.dev/mcp \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"my-agent","version":"1.0"}}}'
Self-Hosted Option
Clone the repository and run the MCP server locally:
git clone https://github.com/MidOSresearch/midos-core
cd midos-core
python -m modules.mcp_server.midos_mcp --http --port 3100
MidOS MCP Data Schema & Taxonomy
MidOS organizes data into a structured hierarchy of vectors, chunks, and plans within the Openclaw Skills ecosystem.
| Component | Format | Description |
|---|---|---|
| Knowledge Chunks | Markdown | 46,000+ segments of validated technical information |
| Vector Embeddings | 3072-d Vectors | Gemini gemini-embedding-001 stored in LanceDB |
| Memory Entries | JSON | Persistent, cross-session user and project metadata |
| Plans | Task Trees | Multi-step goals with status checkpoints and tracking |
| EUREKA Insights | Synthesized Insights | High-level intelligence synthesized from raw chunks |
MidOS MCP Advanced Features
- Hybrid Search (BM25 + Semantic) that significantly outperforms standard vector-only retrieval for technical queries.
- Heartbeat Integration logic for automated context resumption and decision logging at the start of every session.
- Real-time system health monitoring and hive status checks to ensure knowledge pipeline integrity.
- Notification hooks for Discord and webhooks to provide real-time feedback on agent actions within Openclaw Skills workflows.
- Knowledge Preflight caching that checks for duplicate knowledge across 48,000+ chunks in under 20ms.
SKILL.md
Loading
Related Openclaw Skills
Go Language Patterns & Best Practices
A definitive collection of production-ready Go 1.24+ patterns covering concurrency, web services, database access, and testing.
FastAPI Production Patterns
Accelerate backend development with production-ready FastAPI patterns for Pydantic v2, async workflows, and secure deployments.
EvidenceOps: Forensic Media Triage
A professional suite of Openclaw Skills for forensic-grade media triage that ensures data integrity and a verifiable chain of custody for digital evidence.
OpenClaw Smart Agent Hub
An intelligent multi-provider LLM assignment and routing system that optimizes AI agent workflows for cost, latency, and quality.
Next.js App Router Production Patterns
A production-grade implementation guide for Next.js 15 and 16 focusing on async APIs, caching changes, and Turbopack performance.
midos-self-improver
A sophisticated agent learning system that filters, scores, and promotes corrections and patterns into permanent project memory through a quality-gated pipeline.
Featured*
