# luplo **AI memory that survives across sessions, teammates, and vendors.** luplo is a CLI + MCP server + HTTP server for tracking decisions, knowledge, policies, documents, tasks, and QA checks across coding sessions. It is backed by PostgreSQL with full-text search (`tsquery` + glossary expansion) and optional pgvector reranking. Three interfaces share one core: - `lp` — the human-facing CLI (typer). - An **MCP server** on stdio, usable from any [MCP-compatible client](https://modelcontextprotocol.io/clients) (Claude Code, Claude Desktop, Cursor, Zed, custom). - An optional **FastAPI HTTP server** for multi-user deployments. ## Where to start ::::{grid} 1 2 2 2 :gutter: 3 :::{grid-item-card} Quickstart :link: quickstart :link-type: doc Install uv + Postgres, initialise a project, record and recall a decision, wire up an MCP client — in about five minutes. ::: :::{grid-item-card} Concepts :link: concepts/index :link-type: doc Why luplo refuses to be a framework, what honesty-over-coverage means for search, how the twelve-table schema fits together. ::: :::{grid-item-card} Guides :link: guides/index :link-type: doc How-tos for work units, tasks & QA, connecting MCP clients, running the Remote server, operating the worker. ::: :::{grid-item-card} Reference :link: reference/index :link-type: doc CLI flags, MCP tool surface, every configuration variable, and the auto-generated API reference. ::: :::: ## Project - Source — - License — {doc}`AGPL-3.0-or-later ` (CLA coming) - Changelog — {doc}`project/changelog` - Contributing — {doc}`project/contributing` ```{toctree} :hidden: :maxdepth: 1 quickstart concepts/index guides/index reference/index project/index API reference ```