linglong-knowledge

# Tool List > Unified Knowledge Base for Cross-Agent — Providing a Shared Knowledge Base for AI Agents through MCP. ## Project Overview Linglong Knowledge is a unified knowledge base for cross-agent, solving the problem of knowledge islands for agents like OpenClaw, Claude Code, and Codex. It provides knowledge read and write capabilities through the MCP protocol, supporting keyword search, semantic vector search, and hybrid search. ## Architecture ``` Agent（OpenClaw / Claude Code / Codex）──→ Knowledge Store ──→ MCP Server (File + SQLite + sqlite-vec) ``` ``` src/linglong/ ├── core/ # Configuration, Data Models, LLM Client ├── knowledge/ # Knowledge Base Storage (SQLite + sqlite-vec, Review Engine, Lint, Synchronization) ├── mcp/ # MCP Server (10 Tools, stdio / HTTP) └── cli.py # CLI Entry ``` ## Module Status | Module | Status | Description | |------|------|------| | core | ✅ | Configuration Center (`.knowledge.yml`), Shared Data Models, LLM Client | | knowledge | ✅ | SQLite + sqlite-vec Storage, Review Engine, Lint Inspection, OpenClaw Synchronization | | mcp | ✅ | MCP Server (10 Tools), Supporting stdio and HTTP Deployment | > **Note**: ingest (Data Acquisition), reviewer (Article Review), dispatch (Multi-Platform Distribution) have been split into independent projects. ## Quick Start ```bash git clone https://github.com/xinovate/linglong-knowledge.git cd linglong-knowledge # Install Dependencies pip install -e ".[dev]" # Configuration cp .knowledge.example.yml .knowledge.yml # Edit .knowledge.yml as needed # CLI Usage linglong kb lint # Inspect Knowledge Base linglong kb sync # DB↔File Consistency Check linglong kb index --rebuild # Rebuild Index # MCP Server ll-knowledge-mcp # stdio Mode (for Local Agents like Claude Code) ``` ### MCP Registration (Claude Code) ```bash claude mcp add ll-knowledge -- /path/to/.venv/bin/ll-knowledge-mcp ``` ## Configuration The main configuration file is `.knowledge.yml` (search path: `./.knowledge.yml` → `~/.knowledge/config.yaml`). Environment variables with the prefix `KB_` are also supported. ```yaml # .knowledge.yml Example knowledge: wiki_path: ~/knowledge/wiki db_path: ~/knowledge/db/knowledge.db generate_embeddings: true embedding_url: http://localhost:7997 embedding_model: nomic-ai/nomic-embed-text-v1.5 mcp: transport: stdio host: 127.0.0.1 port: 9900 ``` For complete configuration options, refer to [`.knowledge.example.yml`](.knowledge.example.yml) or [Configuration Documentation](docs/config.md). ## Tech Stack - **Python 3.11+** - **Pydantic** — Data Validation and Configuration Management - **SQLite + sqlite-vec** — Local Storage and Vector Search - **FastMCP** — MCP Protocol Implementation - **pytest** — Testing Framework ## Documentation - [Project Overview](docs/PROJECT_OVERVIEW.md) — Version Status and Next Actions - [Architecture Design](docs/architecture.md) — System Architecture and Module Dependencies - [API Documentation](docs/api.md) — MCP Tool Interfaces and Configuration - [Knowledge Base Module](docs/knowledge.md) — Design, Agent Access, and Reference Materials - [Development Guidelines](docs/rules.md) — Code Style, Testing, and Security - [Version Roadmap](docs/roadmap.md) — Evolution Plan and ADR ## License MIT